ammonia borane composites for solid state hydrogen storage ......i ammonia borane composites for...
TRANSCRIPT
Ammonia borane composites for solid state hydrogen
storage and calcium-ammonia solutions in graphite
Atahl Nathanson
This thesis is in accordance with the requirements of the University College London for
the degree of Engineering Doctorate
February 2014
I
Ammonia borane composites for solid state hydrogen
storage and calcium-ammonia solutions in graphite
Submitted for the degree of Doctor of Engineering
February 2014
Abstract
The dual problems of worlds growing population, increasing energy demand and global warming,
necessitate an alternative to fossil fuels. Hydrogen is plentiful and has a high energy density but
storage in high pressure tanks is complex and presents safety concerns. Ammonia borane (AB) is
one of the most promising solid state hydrogen storage materials due to its high releasable
hydrogen content (13.1 wt%), stability in air, and low toxicity. On heating, however, pure AB
releases hydrogen only after long nucleation times and is accompanied by the liberation of
gaseous impurities including borazine and ammonia; additionally, extensive material expansion
and foaming occurs. AB composites with polyethylene oxide, polystyrene and imogolite have
been synthesized. It is concluded that the decomposition of AB is best ameliorated by providing
access to functional groups that catalyse alternative dehydrogenation routes. Lowering the onset
of hydrogen loss to below the melting temperature limits the overall foam and expansion.
The two dimensional confined motion of liquid ammonia in a multi staged ternary calcium
ammonia graphite intercalation compound was studied with respect to temperature. Hopping
diffusion at 300K gives way to rotation below 100K. The dynamics of this confined calcium
ammonia solution are observed as similar to the three dimensional counterpart.
II
I, Atahl Nathanson confirm that the work presented in this thesis is my own. Where information
has been derived from other sources, I confirm that this has been indicated in the thesis
III
Acknowledgements
Firstly, a thank you to my supervisors Stephen Bennington and Neal Skipper for their guidance,
support and the enthusiasm expressed for my scientific endeavours. Both provided distinct and
valuable roles in allowing me to experience both academia and industry.
The team at Cella Energy, Arthur (Thank you for being an invaluable mentor and friend.), Zeynep
(Thanks for all the great food, help with experiments and political awareness.), Stephen (All that
wardrobe advice and unfortunately I still wear shorts with tights.), Tom (We didn’t get shot!),
Dave, Joe, Lucile, Marie, Lindsey (Tango till dawn with me?), Gemma (I’ve still got that duck!),
Anna and Matt (Many thanks, soon we will all be published authors!) and Kevin (Thank you for
starting for my ill car – it’s now in mechanical heaven.).
Now all those at Isis: Tati (where would we be without your bus? Thank you for all those hours
spent waiting for me at the side of the road), Aziz (keep up the music), Jordi (Who will throw
things at me now?), Rowen, Bruno, Konstantin, Davide, Rob, Giovanni, Sev, Roberto (Thank
you, I will come to Milan…soon…very soon!), Vicky (Thank you for all your help and patience.),
Ruth (Thanks for Robbie and all the fabulous gossip.), Ivan (Thanks for all those great coffees
and cakes.) and Joe (Keep peddling.).
Thank you to the F10 crew for support during my write up. Emily, so handy with the hanky and
the prep talks, thank you. Christian, Dave, Paddy, Radhika and Stephen, thanks for the help,
lunches and letting me play my atrocious football. I’m building a team over in Chemistry so
prepare yourselves! Hello and thanks also to the EngD group Nic, Nuru, Will, Szymon, Clyde,
Marcus, Isaac, Liz, Jay and Kathrine, unfortunately dl-poly didn’t make the final cut this time.
Thank you Chris for making the calcium and lithium graphite samples. Spencer and Jaqueline,
without your guidance through Modes the neutron data would still be very raw. Derek, many
thanks for all the science advices! Arthur, gracias for the Sketchup electrospinning picture.
My three keen work experience students Bouyan, Ben and Hannah. Thank you for coming and
being, well, so keen!
Jaqueline (Thank you for your help, you’re always in my heart.), Jake (Thanks for tolerating my
general madness.) and Eloise (your good wishes are always so well timed).
Thank you to all my proof readers; Steve, Neal, Mum, Christian, Joe, Tim and Rhiain.
So to my family, Mum, Dad, Phil, Rhiain, Brighid and Lucien. Thank you for believing in me
and making me feel intelligent. Sarah, Gemma, Jade, Parry, Isaac and Suki, you rock my world,
IV
thanks for letting me live in your home. And last but not least thanks to Robbie my feline bed
warmer.
V
Contents
1 An introduction to hydrogen storage and graphite intercalation compounds ............ 1
Hydrogen....................................................................................................................... 1
Fossil fuels and global warming ........................................................................... 1
A hydrogen economy ............................................................................................ 1
Hydrogen storage targets for vehicles ........................................................................... 4
Current hydrogen storage methods for vehicles............................................................ 4
Storing hydrogen as a gas or liquid ....................................................................... 7
Clathrates .............................................................................................................. 7
Physisorbed hydrogen storage .............................................................................. 8
Chemisorbed hydrogen storage ............................................................................. 9
Alkali metal ammonia graphite intercalation compounds .......................................... 12
2 Materials introduction .............................................................................................. 15
Ammonia borane ......................................................................................................... 15
Structure .............................................................................................................. 15
Ammonia borane synthesis and regeneration ..................................................... 17
The reaction pathway of the dehydrogenation of ammonia borane .................... 17
Modifying the dehydration properties of ammonia borane – literature study ............. 20
Ammonia borane in water ................................................................................... 20
Ammonia borane in alternative solvents ............................................................. 20
Ammonia borane in ionic liquids ........................................................................ 21
Ammonia borane in acidic conditions................................................................. 22
Zeolites and metal organic frameworks .............................................................. 23
Carbon ................................................................................................................. 23
Metal catalysts and ammonia borane .................................................................. 24
Ammonia borane in the presence of polymers .................................................... 24
Materials used and project aims .................................................................................. 25
Polymers ............................................................................................................. 25
VI
Clays ................................................................................................................... 27
Graphite intercalation compounds .............................................................................. 30
3 Experimental methods ............................................................................................. 32
Making samples .......................................................................................................... 32
Electrospinning ................................................................................................... 32
Spinning parameters ............................................................................................ 39
Freeze drying ...................................................................................................... 41
Dry mixing .......................................................................................................... 41
A vapour transport method for intercalation into graphite .......................................... 42
Analysing samples ...................................................................................................... 42
Scanning electron microscopy ............................................................................ 42
Fourier transform infrared spectroscopy ............................................................. 44
Foam tests ........................................................................................................... 45
Thermogravimetric analysis and differential scanning calorimetry .................... 45
Mass spectrometry .............................................................................................. 47
X-ray diffraction ................................................................................................. 53
Nuclear magnetic resonance spectroscopy .......................................................... 54
Introduction to neutron scattering ............................................................................... 55
IRIS ..................................................................................................................... 59
Modes .................................................................................................................. 62
4 Ammonia borane and polyethylene oxide composites ............................................ 63
Electrospinning polyethylene oxide ............................................................................ 63
Electectrospinning polyethylene oxide with ammonia borane ................................... 66
Uncertainty .......................................................................................................... 67
Freeze drying and dry mixing polyethylene oxide and ammonia borane ................... 68
X-ray diffraction of the electrospun and freeze dried polyethylene oxide and ammonia
borane composites ................................................................................................................... 68
Foam tests ................................................................................................................... 70
VII
Ramp heating of the ammonia borane-polyethylene oxide composites and pristine
ammonia borane ...................................................................................................................... 74
Hydrogen release, thermogravimetric analysis and differential scanning
calorimetry .......................................................................................................................... 74
Impurities ............................................................................................................ 76
Conclusions ......................................................................................................... 77
Isothermal heating ....................................................................................................... 77
Hydrogen release under isothermal heating runs ................................................ 77
Non-hydrogen species released form AB under step heating ............................. 79
Conclusions ......................................................................................................... 80
Fourier transform infrared spectroscopy ..................................................................... 85
FTIR of ammonia borane and polyethylene oxide .............................................. 85
FTIR of the AB-PEO fibres ................................................................................ 86
Isothermal heating of 50AB-50PEO studied by FTIR ........................................ 87
Conclusions ......................................................................................................... 89
Nuclear magnetic resonance spectroscopy.................................................................. 91
11B NMR of ammonia borane heated in situ at 85°C .......................................... 91
11B NMR of the 50AB-50PEO fibres heated in situ at 85°C .............................. 93
Conclusions ......................................................................................................... 94
15N NMR ............................................................................................................. 95
Ex situ heating ............................................................................................................. 96
Heating XRD ...................................................................................................... 99
Discussion and conclusions ...................................................................................... 100
Structure of the ammonia borane-polyethylene oxide composites ................... 100
Foaming of ammonia borane and of the ammonia borane-polyethylene oxide
composites ......................................................................................................................... 100
Hydrogen release............................................................................................... 101
Reaction mechanism in the gas stream ............................................................. 102
Further work .............................................................................................................. 103
Structure ............................................................................................................ 103
VIII
Foaming ............................................................................................................ 104
Impurities .......................................................................................................... 104
Reaction Mechanism ......................................................................................... 104
5 Polystyrene and ammonia borane .......................................................................... 105
Electrospinning polystyrene and ammonia borane ................................................... 105
Foam tests on the ammonia borane-polystyrene fibres ............................................. 107
Ramp heating of ammonia borane-polystyrene fibres .............................................. 108
Conclusions and discussion ...................................................................................... 111
Further work .............................................................................................................. 112
6 Ammonia borane and clay ..................................................................................... 113
Scanning electron microscopy .................................................................................. 113
Foam tests of AB-clay composites ............................................................................ 115
Ammonia borane-imogolite: varying the proportion of AB ..................................... 118
X-ray diffraction of AB-IMOG composites .............................................................. 119
Thermogravimetric analysis and differential scanning calorimetry of the ammonia
borane-imogolite composites ................................................................................................ 122
Hydrogen gas stream content .................................................................................... 124
Discussion and conclusions ...................................................................................... 126
Structure ............................................................................................................ 126
Foaming ............................................................................................................ 126
Dehydrogenation ............................................................................................... 127
Further work .............................................................................................................. 128
Structure ............................................................................................................ 128
Foaming ............................................................................................................ 128
Reaction ............................................................................................................ 128
Alternative materials ......................................................................................... 129
7 Neutron spectroscopy studies of calcium-ammonia solutions confined by graphene
sheets. ............................................................................................................................ 130
Synthesis ................................................................................................................... 130
IX
Diffraction ................................................................................................................. 130
Diffusional dynamics ................................................................................................ 133
Conclusions and discussion ...................................................................................... 137
Future work ............................................................................................................... 138
8 Final conclusions ................................................................................................... 139
Structure and foaming of the ammonia borane composites ...................................... 139
Hydrogen release from the ammonia borane composites ......................................... 139
Solid state hydrogen storage ..................................................................................... 140
Calcium graphite intercalates .................................................................................... 140
9 Bibliography .......................................................................................................... 141
X
List of Figures
Figure 1: Ragone plot of the energy storage verses power for various electrochemical energy
conversion systems. The high specific energy of fuel cells can provide a viable alternative to the
internal combustion engine for vehicles [8]. ................................................................................. 2
Figure 2: Proton exchange membrane fuel cells enable the controlled burn of hydrogen to produce
electricity [9]. ................................................................................................................................ 3
Figure 3: The gravimetric and volumetric hydrogen densities of various potential hydrogen
storage materials [3] ...................................................................................................................... 5
Figure 4: Spider plots for several hydrogen storage systems, reversible (top), chemical hydrides
(middle), and sorbent (bottom) with respect to the 2010 DOE targets. The yellow areas show
where improvements in technology and understading are required [13]. ..................................... 6
Figure 5: Van’t Hoff desorption lines for single metal and alloyed metal hydrides from Sandrocks
metal hydride review paper [23]. The box area represents the desired temperature and pressure
range of operation for vehicular applications. The alloys are highly versatile and tuneable. ..... 10
Figure 6: The AA, AB and ABC stacking arrangements of graphite layers. .............................. 13
Figure 7: Possible staging arrangements for graphite intercalation compounds. ........................ 14
Figure 8: The possible dihydrogen bonding arrangement in AB determined by Bowden et al. [44].
A single layer on the ab plane is displayed. Nitrogen atoms are shown in pink and boron atoms
are purple. a) is the orthorhombic phase observed at 200K and b) is the tetragonal phase observed
at 300K. ....................................................................................................................................... 16
Figure 9: AB regeneration with hydrazine and ammonia [52]. .................................................. 17
Figure 10: Thermolysis of ammonia borane determined experimentally by Frueh et al.; both cyclic
and polymeric amido and imido boranes are produced [57] . Slow heating rates (<1°C min-1) do
not produce borazine, medium heating rates (3-10 °C min-1) show borazine during the release of
the second hydrogen equivalent and aggressive heating regimes (>10°C min-1) produce even more
cyclic by-products [59]. .............................................................................................................. 18
Figure 11: The themolysis reaction mechanism of AB (polymeric) proposed by Stowe et al.
showing the induction, nucleation and growth phases [67]. ....................................................... 19
Figure 12: A) Decomposition pathway for AB in glyme suggested by Shaw et al. based on boron
NMR experiments [79]. AB follows that cyclic reaction route. B) From Kim et al. showing how
hydrogen bonding between AB and ether promotes formation of DADB [81]. ......................... 21
XI
Figure 13: The mechanism of acid dehydrogenation of AB proposed Stephens et al. [37]. The
acid abstracts a hydride (a) activating the boron to interact with a neighbouring AB molecule (b)
forming an intermediate. Hydrogen loss follows and an intermediate with a bridging hydrogen is
formed. If the acid to AB ratio is low dehydropolymerization ensues (c). However, if the ratio is
high, a second reaction takes place (d) which catalyses the reaction further (e). ....................... 22
Figure 14: Schematic of the PEO monomer unit. ....................................................................... 25
Figure 15: Schematic of the PS monomer unit. .......................................................................... 27
Figure 16: Single phase electrospinning rig. ............................................................................... 33
Figure 17: Schematic of a scanning electron microscope. .......................................................... 43
Figure 18: AB containing pellets prepared for foam testing. ...................................................... 45
Figure 19: Schematic of a combined DSC/TGA instrument. ..................................................... 46
Figure 20: Schematic of a residual gas analyser (mass spec). .................................................... 47
Figure 21: Savitsky-Golay smoothing, of ammonia (m/z = 17) signal where N = 500. ............. 49
Figure 22: Cumulative integrals of the heating ramps of two pristine AB samples (AB1 and AB2)
tested several months apart. The data were normalised to the hydrogen (m/z = 2) level at 190ºC
and errors in the data are shown as the lighter background. The increase in impurities observed in
AB2 is likely due to the time separation between the two tests. This graph has been plotted to
show care must be taken when comparing impurity levels from different samples. .................. 51
Figure 23: The mass loss (top) of the 75AB-25PEO electrospun fibres when subjected to heat
steps of 150°C, 120°C and 100°C in the combined TGA/DSC (bottom). In the 150°C and 120°C
steps the majority of the hydrogen has been lost before the maximum temperature was reached.
Additionally the samples never reach the set temperature. Therefore this loss data are
inappropriate for analysing the activation energy of AB. ........................................................... 52
Figure 24: The momentum and energy changes associated with elastic and inelastic scattering. 𝑘𝑖
and 𝑘𝑓 are the initial and final wavevector of the neutron respectively. .................................... 56
Figure 25: Geometry of a scattering experiment [165] ............................................................... 57
Figure 26: Schematic of IRIS indirect time of flight neutron spectrometer................................ 59
Figure 27: Schematic of the neutron scattering energy spectrum. .............................................. 60
Figure 28: 3% PEO (2M) in a variety of solvents. ...................................................................... 64
Figure 29: PEO electrospun fibres with solvents H2O, ACN and an equal mix of the two. ....... 65
XII
Figure 30: The viscosity of the AB-PEO in ACN solutions with respect to shear speed. The
viscosity at the high shear limit quoted in Table 14. The addition of AB successively reduces the
viscosity. ..................................................................................................................................... 66
Figure 31: Images of the AB-PEO fibre samples, 24AB-75PEO, 50AB-50PEO and 75AB-25PEO.
As the ratio of AB to PEO increases the fibres become rougher. ............................................... 67
Figure 32: The fluffy fibres produced from the PEO-AB-ACN solution. .................................. 68
Figure 33: X-ray diffraction pattern of PEO as received granules and electrospun fibres. ........ 69
Figure 34: X-ray diffraction pattern of AB, PEO granules and the electrospun fibres collected on
a Phillips X’pert, above, and the Rigaku SmartLab, below. ....................................................... 70
Figure 35: Foaming response of a compressed pellet of AB when subjected to 120°C isothermal
heating ......................................................................................................................................... 71
Figure 36: Foam tests with pellets made from electrospun fibres, freeze dried material, and hand
milled powder with AB contents, 25, 50 and 75wt% ................................................................. 72
Figure 37: Unfurled pellet ........................................................................................................... 73
Figure 38: Representative examples of the relation between initial pellet density and foaming. 73
Figure 39: Combined TGA, DSC and RGA hydrogen signal (m/z=2) for pristine AB and the AB-
PEO electrospun fibres. As the fraction of PEO in the sample increases, the hydrogen release is
activated at earlier temperatures. Mass loss occurs earlier in the AB-PEO composites than in the
pristine AB but the final wt% loss is similar (~50%) in both. In the DSC curves, successive
addition of PEO reduces the depth of the AB melting endotherm. The lower temperature release
of hydrogen is accompanied by an exotherm which reduces the apparent size of the melting
endotherm. .................................................................................................................................. 75
Figure 40: Cumulative integrals of the heating ramps of pristine AB and the AB-PEO composite
fibres. It is noted that raising the amount of PEO in the mix aids the hydrogen release but produces
more borazine. The data were normalised to the hydrogen (m/z = 2) level at 190ºC and errors in
the data are shown as the lighter background. ............................................................................ 78
Figure 41: Hydrogen (m/z=2) trace of the isothermal heating runs of pure AB and the composite
materials. At all temperatures, the dehydrogenation of AB begins earlier and is accelerated as the
PEO fraction of the composite increases. ................................................................................... 81
Figure 42: The borazine signal, m/z=81, from pristine AB and the AB-PEO fibres under
isothermal heating at 80°C, 100°C, 120°C and 150°C. The pristine AB does not release any
borazine at any of the temperatures. This is consistent with literature and the heat ramp data
(Figure 40) where borazine is first observed at 150°C. (The sample is actually several degrees
XIII
cooler than the written value, see experimental chapter 3). None of the AB-PEO fibres release
borazine at 80°C but all give off borazine at 120°C and 150°C. At 100°C the 25AB-75PEO is the
only sample that releases significant borazine. This is consistent with the ramp heating tests where
it is first observed around 100°C in this sample. This suggests the PEO is encouraging borazine
to be released............................................................................................................................... 82
Figure 43: The diborane signal, m/z = 27, released by pure AB and the AB-PEO fibres under
isothermal heating of 80°C, 100°C, 120°C and 150°C. No diborane was detected at 80°C and
100°C. At 120°C both the pristine AB and the AB-PEO fibres show borazine. At 150°C the
diborane level are higher in the AB-PEO fibres than in the AB. This suggests PEO encourages
the formation of diborane at high temperatures. ......................................................................... 83
Figure 44: The ammonia signal, m/z= 17, released by pristine AB and the AB-PEO fibres under
isothermal heating of 80°C, 100°C, 120°C and 150°C. The ammonia levels are similar in the pure
AB and the fibres, but ammonia is released sooner in the fibres. At 80°C AB does not release any
ammonia, but then it doesn’t release any hydrogen either. This suggests PEO does have a
significant effect on the ammonia release. .................................................................................. 84
Figure 45: The FTIR spectrum of as received AB, electrospun PEO fibres and 50AB-50PEO
fibres. Stretches are identified from literature. [179]–[183]. ...................................................... 86
Figure 47: The FTIR spectra of AB, PEO and the three electrospun fibres. As the PEO content in
the fibres increases there is a change in intensity (grey highlights), a shift to higher wave number.
Additionally there is splitting of the PEO C-O and C-H stretching peaks (dashed line). This
supports the previous suggestion that the dihydrogen bonding of AB is being replaced by
hydrogen bonding between AB and PEO. .................................................................................. 88
Figure 48: FTIR of the heated samples. New peaks that appear during the heating are labelled.
.................................................................................................................................................... 90
Figure 49: 11B NMR of pure AB at 85°C. ................................................................................... 91
Figure 50: 11B NMR of 50AB-50PEO material heated at 85°C for several hours. ..................... 93
Figure 51: The 11B NMR studies of AB dehydrogenation at 85°C. The pristine AB superimposed
on the 50AB-50PEO fibre sample. The AB in the fibres reacts faster, the BH4 group has a broader
peak and B-O bonds occur. ......................................................................................................... 95
Figure 52: 15N NMR of pure AB and 50AB-50PEO electrospun sample before and after 6 hours
heating 85°C when PAB is observed. ......................................................................................... 96
Figure 53: The changing surface of the 50AB-50PEO electrospun fibres at 85°C. .................... 98
Figure 54: Higher resolution image of pockmarked ................................................................... 99
XIV
Figure 55: X-ray diffraction pattern of the 50AB-50PEO electrospun fibres heated ex situ over 3
hours. The peaks associated with AB and PEO disappear and are replaced with an amorphous
hump. A hump at 22° has previously been associated to PAB. .................................................. 99
Figure 56: SEM image of PS and 50AB:50PS electrospun fibres. ........................................... 107
Figure 57: Foam testing AB-PS pellets made from the electrospun fibres. Foaming and residue
on the test tube wall is inversely related to the PS content. Images of the final pellets once removed
from the test tube have also been inserted. ............................................................................... 108
Figure 58: TGA trace of the mass loss of the AB-PS fibres when subject to a 2°C/min temperature
ramp between room temperature to 200°C. .............................................................................. 109
Figure 59: DSC traces of the AB-PS electrospun fibres under ramp heating rate of 2°C/min. 109
Figure 60: The hydrogen and impurity trances for the AB-PS fibres. When combined with PS,
AB releases hydrogen at lower temperatures than when in its pristine state. The PS does not appear
to have a significant effect on the impurity levels. ................................................................... 110
Figure 61: SEM images of the four AB-clay composites. In the case of IMOG and LAP, the
significant morphology change between the as received clay and the AB-clay composite, suggests
intimate mixing of the clay and AB has been achieved. ........................................................... 114
Figure 62: AB:clay 1:1 ratio pellets post foam testing. ............................................................ 115
Figure 63: The LAP and IMOG AB-clay pellets post foam testing. The AB-LAP pellet instantly
crumbles while the AB-IMOG remains whole. ........................................................................ 116
Figure 64: The change in mass, density and volume of the AB-clay pellets as a percentage of the
initial dimensions post foam testing.......................................................................................... 116
Figure 65: Volume expansion against initial density of the AB-clay composites. The AB-IMOG
is the only sample that suggests a positive correlation. ............................................................ 117
Figure 66: AB-IMOG pellets after foam testing. ...................................................................... 118
Figure 67: SEM image of the 1AB-3IMOG, 2AB-2IMOG and 3AB-1IMOG freeze dried
composites. As the proportion of AB increases the visible free volume decreases. ................. 119
Figure 68: The two main AB tetragonal peaks for pristine AB and IMOG. The relative
displacement is likely due to the difference of the height of the sample as shown in the
displacement of the sample holder peak. .................................................................................. 119
Figure 69: The AB peaks in the composites are observed at the same 2 theta position............ 120
Figure 70: XRD pattern of the AB IMOG freeze dried composites. ........................................ 121
XV
Figure 71: The freeze drying process used to prepare the AB-IMOG composite leaves the IMOG
tubes dispersed within the AB. ................................................................................................. 121
Figure 72: Hydrogen release profile of the AB-IMOG composites compared to pure AB under a
standard ramp heating. As the IMOG content increases the pre-peak increase and the release of
the 1st and 2nd hydrogen equivalent (H2 equiv.) ........................................................................ 122
Figure 73: TGA and DCS traces for the 1AB-3IMOG, 2AB-2IMOG and 3AB-1IMOG freeze
dried composites under standard ramp heating. ........................................................................ 123
Figure 74: Cumulative integrals of the mass spec traces of hydrogen and impurities detected
under the heating ramp. The data are normalised to the hydrogen (m/z = 2) level at 190ºC and
errors in the data are shown as the lighter background. As the IMOG content is raised more
ammonia is released but borazine and diborane are decreased. ................................................ 125
Figure 75: Hypothetical reaction scheme for the initiation (prepeak) of hydrogen release from AB
by the hydroxyl groups on the IMOG surface. ......................................................................... 127
Figure 76: The neutron diffraction pattern pre ammoniation (purple line) shows both Ca GIC and
graphite. After ammoniation (long red trace) a new multistaged ternary phase Ca-NH3 GIC was
created. The final Ca-NH3 GIC sample includes stages 2, 3 and 4 labelled GIC2, GIC3 and GIC4
respectively. During the NH3 intercalation unidentified high stages (marked *) form before the
low stages. ................................................................................................................................. 131
Figure 77: The layers of the Ca-NH3 GIC contract when the temperature decreases from 300K to
2K. ............................................................................................................................................. 132
Figure 78: In graph A the total neutron scattering spectrum of the Ca-NH3 GIC is plotted as a
function of Q for a range of representative temperatures. The fit to the data are good as
demonstrated by a small residual. The spectra broaden and the peak height drops as the
temperature increases. In graph B the elastic incoherent structure factor, EISF, ratio of the elastic
scattering over the total scattering for all Q, against temperature is depicted. With increasing
temperature the shift from elastic to inelastic scattering is due to the protons becoming more active
and indicates that greater motion is available to the ammonia. Considering the sharp drops in the
trace of the right-hand graph, we can preliminarily define the motional modes occurring. ..... 133
Figure 79: The FWHM of the Lorentzian fits to the data at 300 and 250K, previously divided into
17 groups to depict the changes as a function of Q2. The Singwi-Sjölander jump model provides
the best fit to the data describing an exponential distribution of jump lengths and frequency around
the mean values. At 300K this corresponds to a jump distance of 3.2±0.1Å with a frequency of
19ps and at 250K, this corresponds to a jump distance of 2.88±0.2Å with a frequency of 28 ps.
.................................................................................................................................................. 134
XVI
Figure 80: Molecular graphics snap-shot of a single layer of composition CaC24(NH3)2 showing
the jump lengths expected (Unpublished work from Neal Skipper). ........................................ 135
Figure 81: In graph A the FWHM are shown with respect to Q2 at 200K, 150K and 100K. The
linear nature of the fit suggests rotational modes with higher energy at higher temperatures. In
graph B an Arrhenius plot of the FWHM obtained yields the activation energy off the rotation as
696 Jmol-1. ................................................................................................................................. 135
Figure 82: The QENS spectrum before ammoniation, during the experiment and after the
ammonia was pumped off, the change in intensity demonstrates the partial reversibility of
ammonia intercalation. .............................................................................................................. 136
XVII
List of Tables
Table 1: The revised US DOE hydrogen storage targets for automobile applications [11]. ......... 4
Table 2: The specific and volumetric hydrogen capacities of liquid and gaseous hydrogen ........ 7
Table 3: Selected metal hydrides, hydrogen content and decomposition temperatures. ............. 12
Table 4: The hydrogen content of ammonia borane [37]. ........................................................... 15
Table 5: Electron distribution in the AB molecule. .................................................................... 15
Table 6: Bond distances in ammonia borane, theoretical and experimental sources. ................. 16
Table 7: Solvents used in practical work, boiling point and vapour pressure. ............................ 37
Table 8. The mass number assignment of the ionised species observed in the mass spec.......... 47
Table 9: Neutrons according to temperature [164] ..................................................................... 57
Table 10: Scattering lengths and cross sections of the nuclei observed in experiments in this thesis.
.................................................................................................................................................... 58
Table 11: Jump diffusion models, l is the jump length and 𝝉 is the average translational residence
time of the diffusing particle. ...................................................................................................... 61
Table 12: An illustration of the electrospinning parameters tested, to determine the ideal
conditions to produce well defined PEO fibres from PEO in water solutions. ........................... 63
Table 13: Electrospinning parameters of PEO (2M) solutions in H2O, in ACN and in a 50:50 mix
.................................................................................................................................................... 65
Table 14: Processing parameters used for the electrospinning the AB-PEO in ACN solutions. 67
Table 15: The stretching frequencies observed in the FTIR that correspond to B-O bonds ....... 89
Table 16: NMR peak assignments. ............................................................................................. 92
Table 17: The spinning conditions tested for the PS in a 1:1 ratio of THF:DMF to determine the
ideal system parameters before adding AB. .............................................................................. 105
Table 18: The spinning conditions tested for the AB-PS single phase solution to determine the
ideal system parameters to produce stable spinning conditions and dry fibres. ....................... 106
Table 19: The neutron diffraction pattern shows the Ca-NH3 GIC to have multiple stacking
arrangements. Stage 2, 3 and 4 are present but no stage 1 is observed. This is due to the low NH3
pressure of 3bar which limits the amount of NH3 in the galleries. ............................................ 131
XVIII
Table 20: The FWHM, rotational dynamic correlation times, 𝜏𝑟𝑜𝑡, and rotational diffusion
constant, Drot, calculated at temperatures of 100K, 150K and 200K. ....................................... 136
1
1 An introduction to hydrogen storage and graphite
intercalation compounds
Hydrogen
Hydrogen is the smallest and most abundant (more than 90%) element in the universe [1]. It has
a high energy density of 39.1kWh kg-1 about three times that of diesel making it an ideal energy
carrier [2]. It is colourless, odourless, and nontoxic. Hydrogen is highly flammable and highly
versatile, it can be burnt in a fuel cell, a normal combustion engine or as a component of the
natural gas on a kitchen hob. The only product from this combustion is water. This makes
hydrogen an ideal energy carrier.
Fossil fuels and global warming
Each year, human activities cause 3×1012 kg of carbon to be released into the atmosphere in the
form of CO2. This raises the CO2 concentration by roughly 0.4% annually [3]. The atmospheric
CO2 level is tied closely to the global temperature and warming of the Earth’s surface over the
last 100 years is already causing changes in climate incompatible with sustaining human life.
Fossil fuels are essentially chemical energy carriers, where the sun’s energy has been stored away
by the compression of organic matter over millennia. The time scales required for the creation of
coal and oil mean that as well as being polluting, fossil fuels are a quickly diminishing finite
resource.
It has been shown that energy, from fossil fuels, has enabled the dramatic rise in living standards
we are now enjoying in so many parts of the world [4]. Ending the worldwide use of fossil fuels
is vital for sustainability but society relies heavily on cheap energy and ‘pulling the plug’ is not
viable option. Sustainable development requires renewable energy sources with versatile energy
storage and energy carriers.
A hydrogen economy
It has long been suggested that our current fossil fuel economy could be replaced with a hydrogen
economy [5]. Such a scenario would require efficient production, storage, transportation and
usage of hydrogen. Hydrogen is CO2 neutral. The chemical energy in the hydrogen covalent bond
can be accessed through its cycle with oxygen [6].
2
𝐻2 +1
2𝑂2 ↔ 𝐻2𝑂 𝛥𝐻 = 121 𝑘𝐽 𝑔
−1 ( 1)
Currently, most of the world’s hydrogen is produced through steam reforming of hydrocarbons,
a costly and environmentally dirty method that releases CO2 [6].
𝐶𝐻4 + 2𝐻2𝑂 → 4𝐻2 + 𝐶𝑂2 ( 2)
The most ideal hydrogen production method is from renewable energy produced by hydro, wind
or solar cell power. The electricity produced can be used to electrolyse water to hydrogen and
oxygen, see equation 1. The efficiency of water splitting can reach 83% but practically is in the
range of 70 to 75%. An important consideration is the purity of the water that can be supplied to
the electrolysis cell; impure water can severely impact the lifetime of the cell.
The recent developments of fuel cells over a range of size scales (1 to 100kW) through which
hydrogen can be controllably burnt, facilitate the use for hydrogen to power a range of objects
from mobile phones to vehicles. As 30% of global energy use is for transport [7] hydrogen,
battery or hybrid energy storage systems would go a long way to reducing the global carbon
output.
Figure 1: Ragone plot of the energy storage verses power for various electrochemical energy conversion
systems. The high specific energy of fuel cells can provide a viable alternative to the internal combustion
engine for vehicles [8].
The specific energy of a fuel cell comparable to that of an internal combustion engine, and it
outperforms capacitors and batteries [8]. This means that a vehicle with a fuel cell engine systems
can be lighter than a vehicle powered by a battery system. Capacitors can provide more power
3
per unit mass then fuel cells but lack the necessary specific energy to be useful. In this way fuel
cells provide a very viable alternative to the internal combustion engine for powering vehicles.
A fuel cell can convert hydrogen to electrical current with an efficiency of 60%. Oxygen
dissociates at the cathode, while hydrogen is split at the anode and they combine to form water
and electrons see Figure 2. The electrons are used to run an electronic motor and power the
vehicle.
Figure 2: Proton exchange membrane fuel cells enable the controlled burn of hydrogen to produce
electricity [9].
Hydrogen distribution faces many challenges. Piping hydrogen requires 4.5 times more energy
than natural gas and over long distances becomes expensive [10]. However, the boil off that would
occur when shipping liquid hydrogen makes piping the most suitable option available.
Because hydrogen occupies a large volume under ambient conditions, containing it in
concentrated volumes is problematic. Additionally, being a tiny molecule, many materials are
hydrogen permeable; particularly hydrogen can diffuse into steel causing embrittlement and
cracks. Much effort and many solutions have been proposed as to the best ways to store hydrogen
until it can be used and some of these are discussed in the following hydrogen storage section.
A word of warning; there are suggestions that hydrogen economy could have unforeseen effects
on the environment [6]. Increasing usage of hydrogen would lead to large amounts of hydrogen
being released into the atmosphere primarily from leaks. Hydrogen could dissociate and react
with oxygen increasing the water levels in the stratosphere that could disrupt the polar region
ozone chemistry. Leaking hydrogen could also result in high levels of greenhouse gasses in the
atmosphere as in the troposphere hydrogen radicals are an essential ingredient in methane
forming. There is still much debate on how much hydrogen ultimately would be released and
4
how exactly this would impact the globe, indeed as fossil fuels use diminishes the current release
of hydrogen from incomplete combustion would also reduce.
Hydrogen storage targets for vehicles
The main challenge with hydrogen storage in a vehicle is achieving a sufficient volumetric density
in the fuel tank to realise a 500km (300 mile) driving range [11] see Table 1. Meanwhile the
storage system cost, weight, and refuelling time must be kept low. Also the efficiency and
performance over many refuelling cycles of the system must remain high.
An ideal system would work at ambient temperature and pressure to minimise cost. Also it should
be environmentally friendly (be a closed cycle that does not release by-products that negatively
impact the environment). It is also important to supply a clean hydrogen stream to the fuel cell to
maintain its capacity and lifetime. The tank and/or material that fills it, should be cheap to produce
and easy to recycle. Additionally the system must be safe and convenient to use.
Table 1: The revised US DOE hydrogen storage targets for automobile applications [11].
Year 2010 2015 Ultimate
Gravimetric capacity wt % 4.5 5.5 7.5
Wh g-1 1.5 1.8 2.5
Volumetric capacity g l-1 28 40 70
kWh l-1 0.9 1.3 2.3
Fill time min 4.2 3.3 2.5
Current hydrogen storage methods for vehicles
There are many distinct methods for storing hydrogen. In Figure 3 hydrogen content by mass and
volume of various potential hydrogen storage materials are plotted with respect to the DOE 2010
and 2015 targets [3]. These targets have now been revised to the less ambitious values in Table
1.
5
Figure 3: The gravimetric and volumetric hydrogen densities of various potential hydrogen storage
materials [3]
Conceptually, the simplest hydrogen storage method is as a pressurised gas. High pressure tanks
are well developed and used to power various vehicles such as cars, mopeds and buses [3], [12].
To raise the volumetric hydrogen content, the gas can be liquefied by cooling to below 20K.
Maintaining such low temperatures is at best technologically challenging and limits the usefulness
of liquid hydrogen storage. To reduce the necessity of high pressures and cryogenic temperatures,
highly porous materials with active sites can be used to adsorb and store hydrogen, like bats
hanging in interconnecting tunnels. Promising physisorbed systems include carbon, zeolites and
metal organic frameworks, however, they cannot meet the DOE requirements at RTP. In the case
of chemical and metal hydrides the hydrogen molecule is dissociated and the hydrogen atoms
bonded into a new material. Chemical hydrides have high hydrogen contents, however, they tend
to be either too stable – where the hydrogen is difficult to extract - or conversely, too unstable -
where the hydrogen is difficult to incorporate. Operating under high temperatures and with
catalysts is necessary to render them useful. Here it becomes clear that in addition to volume and
mass hydrogen content, hydrogen storage materials can also be characterised by their operation
temperature and pressure. Additional requirements such as the fill time, dehydration time,
hydrogen purity and cost must also be considered to determine if a material could be an
appropriate hydrogen storage system. Spider plots, where the characteristics of a hydrogen storage
material are rated between 0 and 100 depending on how closely they match the DOE requirements
demonstrate, at a glance, how ‘good’ a particular material is. In Figure 4 the adsorbent systems
are shown as more desirable then the chemical and metal hydrides. A deeper discussion of the
various methods of hydrogen storage follows.
Revised Ultimate system
targets
6
Figure 4: Spider plots for several hydrogen storage systems, reversible (top), chemical hydrides (middle),
and sorbent (bottom) with respect to the 2010 DOE targets. The yellow areas show where improvements
in technology and understading are required [13].
7
Storing hydrogen as a gas or liquid
At ambient conditions hydrogen is a gas with a very low density 0.09kg m-1. To achieve a
significant energy density, suitable for powering a vehicle, the hydrogen has to be compressed.
Containing hydrogen is difficult, high pressures are required and being small it can diffuse into
most materials. Also at high pressure the compression has an energy penalty (~15% for a 700bar
tank). Tanks are lined with barrier material to limit the hydrogen diffusion such as aluminium
oxide or austenitic steel [4], [14]. An outer layer of tightly wound carbon fibre allows high internal
pressures. There are two types of gas cylinder for on board hydrogen storage currently available
on the market; the low pressure, 350bar and the high pressure system at 700bar. Even at high
pressures 5kg of hydrogen can require around 220l of space, in a car the space is usually
scavenged from the passenger seats and boot. Even though the high pressure storage cannot meet
the DOE targets, hydrogen tank systems are currently the best method for on board hydrogen
storage. The refuelling times are fast (5min), the systems are already well developed in a range
of vehicles. Gas tanks are particularly ideal in busses where storage space is plentiful, high speeds
are not typical and the driving range can be short.
Storing hydrogen in its liquid form increases the energy density by around a third while avoiding
the high pressures associated with gas storage see Table 2. Hydrogen boils at 20K so a liquid
system has to be heavily insulated from the surroundings. Liquefying is also expensive costing
30% of the total energy of the hydrogen. In all existing systems boil off (0.3 - 3% per day) is a
significant problem and tanks are designed currently to withstand pressure as well as the
cryogenic temperatures. A release valve is also a must. Toyota have recently developed a system
with 6.1wt%, and 800km range but currently it has too many boil off issues.
Table 2: The specific and volumetric hydrogen capacities of liquid and gaseous hydrogen
State Temperature,
K
Pressure,
bar
Hydrogen
content, wt%
Density, kg
m-3
Energy density,
kWh kg-1
Gas 298 350 3 23.3 0.9
Gas 298 700 4.5 39.3 1.8
Liquid 20 2 6.1 67.67 2.3
Clathrates
Clathrates are frameworks of water molecules stabilised by a guest non polar molecule such as
gases or small organic molecules [15]. They generally resemble ice, and like ice no new bonds
are formed between the molecules rather, a phase change has occurred. Hydrogen was initially
thought too small to support a clathrate however, in 1999 hydrogen was observed to form
clathrates at high pressure (1.5GPa). More recently up 5.3mass% hydrogen has been encased in
8
a type II clathrate (136 water molecules) [14]. However, further studies showed that added actual
number of guest hydrogen molecules varies with pressure and temperature. Tetrahydrofuran can
be used to stabilised the framework to allow the hydrogen to remain in the cages at conditions
closer to ambient however further work is required before such a system could be employed.
Physisorbed hydrogen storage
In physisorbed systems, the hydrogen maintains its molecular form and associates weakly to the
surface of the support material [14]. A high surface area is necessary to increase the number of
binding sites per unit mass. The enthalpy of adsorption is weakly exothermic of 1-10kJ mol-1.
Physisorbed systems have an advantage over chemically bound hydrogen as far less heat is
evolved on hydrogenation. Additionally, physisorbed materials have a long life time as the
material structure is unaffected by the hydrogen cycling through. There are several candidates for
hydrogen storage including, carbon, polymers zeolites and metal organic frameworks (MOFs)
While diamond and graphite are solid, carbon allotypes can have very high surface area [1], [14].
These allotypes include activated carbon, templated carbon and nanotubes. The binding energy
of hydrogen to carbon (<10 kJ/mol) is too weak for ambient temperatures; 0.6wt% was achieved
for activated carbon at ambient conditions. However, on cooling (to 77K) and raising the pressure
(to 20bar) up to 5wt% was recorded [16]. One of the problems with activated carbon is that it is
disordered and the porosity was difficult to control, regular porous carbon materials can however
be achieved by templating. Chemical vapour deposition of carbon on to an ordered zeolite
produced a material with small even pores capable of storing up to 6.9wt% hydrogen at 77K and
20bar [17]. This study also demonstrated that if the pores around 1nm in diameter were the most
effective at holding hydrogen as more hydrogen to carbon interactions occurred. Nanotubes while
attractive could only reach a capacity of around 2wt% because of the inaccessibility of binding
sites [18].
Metal-Organic Frameworks (MOFs) are 3 dimensional networks formed of metal centres
coordinated by organic ligands [1]. MOFs have very low densities due to their high porosity. The
metal centres and ligand types can be exchanged so a vast variety of different MOFs can be
created. Hydrogen storage is highly reversible with fast kinetics; however, a high hydrogen
capacity can only be achieved at low temperature and high pressure. MOFs are also difficult to
manufacture; in MOF-5, ~5wt% storage was achieved at 50 bar and 77K [19] but other groups
with a different production method only achieved 1.6wt% [20]. One of the highest capacity
hydrogen containing MOFs is MOF-177 which achieved 7wt% at 77K [21]. Of the many types
of MOFs studied, the key to high adsorption was linked to high internal surface area. Again small
pores were helpful especially at low pressures.
9
Zeolites are three dimensional silicate structures, of the general formula 𝑀𝑥 𝑧⁄ [(𝐴𝑙𝑂2)𝑥(𝑆𝑖𝑂2)𝑦] ∙
𝑚𝐻2𝑂 where M is an exchangeable cation [1]. Every tetragonal silicate that is substituted by an
alumina, requires also an additionally metal to balance the charge. Zeolites are very flexible and
diverse due to the multiple shapes that can be produced and metals that can be included in the
framework. Zeolites are highly porous with interconnected channels. The hydrogen can easily
penetrate and clings with the charged metals. At low hydrogen loading pressures the number of
metal centres determines the loading that can be achieved. These however fill up quickly and at
high pressure the hydrogen interacts with the zeolite wall, where the pore size determines the
amount of hydrogen that can be supported. Low temperatures are also helpful. Zeolite X can hold
0.1wt% hydrogen at RTP but increases to 2.2wt% on cooling to 77K.
Polymers of intrinsic porosity (PIMS) made of light atoms, are highly porous and have a high
specific area [22]. PIMs are easy to manufacture and have hydrogen adsorption capacities
comparable to zeolites and MOFs; 1.7wt% was absorbed below 10bar in a cyclotricatechylene
PIM. Again hydrogen has a preference for small pores.
Chemisorbed hydrogen storage
1.3.4.1 Metal hydrides
Most metals will react with hydrogen to form hydrides. There are many varieties but all have with
a general formula MHn where n=1, 2, 3 etc. depending on the valance of the metal. Hydrides can
be heavy but have a high energy content per unit volume. They can be formed by supplying
hydrogen at a constant pressure to the metal; the hydrogen will penetrate into the surface and
diffuse through until the metal has fully reacted. Hydrides are reversible, but the exothermic
nature of most metal hydrides makes on-board filling a difficult challenge.
𝑀 +𝑛
2𝐻2 ↔ 𝑀𝐻𝑛 +𝑄, ( 3)
where M is the metal and Q is the heat. In addition to the heat produced, the metal crystal has to
expand to accept the hydrogen and the expansion is hard to control as there are hysteresis effects.
After repeat cycles of hydrogenation and dehydrogenation the metal can become pulverized and
begins to loose capacity.
Most single metal hydrides are either too stable or too unstable to be used for on board hydrogen
storage. In fact, vanadium hydride it the only material that can operate in the range between 1–
10atm and 0–100°C see figure X. This has led to the synthesis and analysis of hundreds of alloys
of different ratio of stable and unstable metal hydrides to tune the dissociation energy to a level
suitable for hydrogen storage applications.
10
Figure 5: Van’t Hoff desorption lines for single metal and alloyed metal hydrides from Sandrocks metal
hydride review paper [23]. The box area represents the desired temperature and pressure range of
operation for vehicular applications. The alloys are highly versatile and tuneable.
While the alloys represent highly versatile and tunable hydrogen storage material their
applicability is limited by the volume expansion and heat evolved during hydration. Further most
of the solids have a reversible hydrogen content of maximum 3wt%. Further work is required i.e.
more different alloys need to be tested, more synthesis methods need to be explored and more
systems engineering to cope with the heat expounded during the reaction.
1.3.4.2 Complex hydrides:
Complex hydrides are materials where hydrogen containing group is attached to a metal. There
are three main types; the alanates (metal with AlH4), borohydrides, (metal with BH4) and imides
and amides (metal with NH or NH2 groups). In the borohydrides and alanates the hydrogen is
hydridic (H-) while in the amides it is protic (H+). Complex hydrides decompose in steps, the first
at low temperatures and the second and or third at higher temperatures.
Alanates have many promising characteristics they are light, cheap and have low toxicity. Also,
their hydrogen content is high. However, alanates also have a very high kinetic barrier to
dehydrogenation see Table 3. NaAlH4 is the most promising; it is the complex hydride that is
closest to fulfilling the DOE’s requirements for on bored hydrogen storage. Dehydrogenation is
slow but at just over 200°C three of the four hydrogen molecules can be released. The addition of
catalysts can improve this such as Ce can improve this. Now high temperature (200°C) PEM fuels
cells are available and systems involving doped NaAlH4 these use approximately 3kg of material
that releases 3.6wt% of stored hydrogen.
Borohydrides have very high wt% hydrogen contents compared to alanates because of the small
size of the boron compared to the aluminium. Most are extremely stable and only decompose at
high temperatures, in fact the temperature required to release hydrogen increases with metal size,
11
see Table 3. LiBH4 is has therefore the best characteristics. Borohydrides also suffer from slow
kinetics, poor reversibility and sometimes release diborane in the hydrogen stream.
Imides and amides are very interesting as they contain both metallic and non-metallic entities.
Initially seen as a breakthrough, soon it became apparent that they suffer from the poor
reversibility, slow kinetics and high operating temperatures. An excellent review has been written
by Ichikawa and can be found in the Handbook of Hydrogen Storage [1]
1.3.4.3 Complex transition metal hydrides
Complex transition metal hydrides (CTMHs) are complex metal hydrides that contain transition
metals [6]. Unfortunately the presence of the transition metal makes the complex very heavy and
the hydrogen storage capacities become uncomfortably low. Additionally CTMHs require
complex and expensive high pressure synthesis methods (500 MPa). This is compounded by the
high temperatures required for hydrogen desorption (Table 3) makes them unsuitable as on-board
hydrogen storage materials.
12
Table 3: Selected metal hydrides, hydrogen content and decomposition temperatures.
Borohydrides Hydrogen content, wt% Start of decomposition
LiBH4 18.5 380
NaBH4 10.8 400
KBH4 7.4 500
Mg(BH4)2 13.7 260-280
Ca(BH4)2 9.6 350
Al(BH4)2 16.8 20
Be(BH4)2 20.7 --
Alanates
LiAlH4 7.9 170
NaAlH4 5.6 230
KAlH4 5.7 >300
MgAlH4 9.3 110-130
Ca(AlH4)2 5.9 80
LiMg(AlH4)2 9.4 170
Complex transition metal hydrides
Mg3MnH7 5.2 280
Mg2FeH6 5.5 320
Mg2CoH5 4.5 280
Alkali metal ammonia graphite intercalation compounds
The electron configuration of carbon is 1s2 2s2 2p2. This initially suggests that carbon can only
form two covalent bonds, however, these orbitals can hybridise. In graphite the hybridisation
leads to a 1s2 2sp3 2p1 configuration. The 2sp orbitals are arranged in a plane at 120° to each other
and the 2p orbital is vertical. This enables the carbon atoms to covalently bond into a planar
hexagonal array called graphene [24]. These graphene sheets interact with each other by
combining of the vertical p orbitals into a π bond.
The electrons can move between π bonds and enables graphite to conduct electricity. As the π
bonds are delocalised the graphene sheets can slide over each other giving graphite lubricating
properties. Additionally, a fibre can be formed from graphite that can be woven into a cloth and
this, when impregnated with resin, has been used in applications as diverse as boat hulls and bullet
proof vests. Graphite fibre has also been used in replacement tendons and ligaments because of
its high strength. The layers of graphite are only weakly bonded to each other and as such
secondary species, such as lithium, can be intercalated; this is the property that allows graphite to
be used as battery electrodes [25]. Graphite is the softest and most stable of all the polymorphs of
carbon [24], [26]. It resembles a black opaque rock with a silvery sheen that flakes at the edges.
13
The carbon atoms are covalently bonded together in a hexagonal planar arrangement with a
carbon-carbon bond length of 1.42Å. These graphene sheets are then stacked one above the other
and each sheet is shifted slightly with respect to the one below to a more energy favourable
configuration. There are two main stacking arrangements; the ABAB, where every second layer
has a basal shift of 1.42Å and the ABCABC, where the second and third layers are shifted
respectively by 1.42Å and 2.84 Å with respect to the first layer (figure 1) [27], [28]. Natural
pristine graphite generally follows an ABAB pattern. In either case the spacing between the layers
is controlled by van der Waals interactions which set the graphene layers 3.35Å apart [28].
Figure 6: The AA, AB and ABC stacking arrangements of graphite layers.
The interlayer spacing in graphite provides an ideal hosting environment for group I and II alkali
metals, the graphene layers move apart to admit the metal and accept the charge from the valance
electrons [24], [26]. Many varieties of metal binary graphite intercalation compounds (GICs) are
possible with a range of interesting structural and electronic properties. GICs are unique in that
they undergo staging to minimise the strain on the structure due to the expansion; instead of a
random distribution, the intercalates are placed between specific, equidistant graphene layers [26],
[29], [30]. The stage is the number of graphene layers present between an intercalate layer. The
first GIC was made in 1841 [31] and the first metal-ammonia GIC was made in 1951 by Rüdorff
and Schulze [32].
1.42 0.00 1.42 and 2.84
AA AB ABC
Relative basal shift (Ǻ)
14
Figure 7: Possible staging arrangements for graphite intercalation compounds.
Ternary GICs, where a second complementary intercalate species is admitted in addition to the
alkali metal are also possible, common entities include H2, N2, CH4, and NH3 [33]. Ternary metal
GICs can only be made if a portion of virgin graphite is present as the secondary molecule
typically displaces a portion of the metal in creating space for itself. Secondary species
intercalation is swift, under 20 hours. This indicates a very high mobility in a two dimensional
environment particularly interesting to study as it is coupled with the structural reorganization of
the graphite planes [26]. While the intercalation of the primary metal is reversible, it has been
shown the secondary species cannot always be fully extracted; when deintercalating NH3
intercalated K-GIC around two NH3 per K remain trapped in the structure [34]. It is suggested
that the residual ammonia is attached to metal ions in defect sites [35] or ensnared in islands [26].
Stage 1 Stage 2 Stage 3 Stage 4
Carbon Intercalate
15
2 Materials introduction
Ammonia borane
Ammonia borane (AB) is one of the most promising complex hydrides for hydrogen storage: it
contains 19.6 wt% hydrogen, two thirds of which can be released below 200°C [2], [36], [37]. It
is a white solid that is relatively stable in air and moisture and, recently, successful recycling of
dehydrogenated AB has also been demonstrated [38]. Its hydrogen content by volume and mass
are well within the ultimate system targets as set by the DOE. With these properties, AB has the
potential to meet the DoE targets for an on-board hydrogen storage material [11], [37]. It does,
however, possess some critical disadvantages: its hydrogen release is preceded by a long
nucleation time, and the decomposition produces foam and generates poisonous volatile by-
products including ammonia and borazine [2].
Table 4: The hydrogen content of ammonia borane [37].
Hydrogen Content wt% Gravimetric g/kg Volumetric g/L
Ammonia borane 19.6 190 100-140
13.1 130 65-80
Structure
AB is formed from a borane and ammonia group connected by a B-N dative covalent bond; the
lone pair of electrons on the nitrogen is donated into the vacant p-orbital on the boron [39].
𝐻3δ+ −𝑁: → 𝐵 − 𝐻3
δ− ( 4)
The Pauling electronegativity values for hydrogen, boron and nitrogen are 2.2, 2.04 and 3.04
respectively, so the electrons are unevenly distributed towards the nitrogen side of the molecule.
This induces a dipole moment across AB of 5.22D [40], [41] where the B–Hδ- and the N–Hδ+
(Table 5).
Table 5: Electron distribution in the AB molecule.
Atom Mulliken charge, e- [39] Bader charge analysis, e- [42]
N −0.91 --
B −0.29 --
HB −0.04/-0.06 -0.44
HN 0.45 0.62
The partial charges on the boron and nitrogen side of AB pull the molecules close together. The
hydrogen atoms on alternate AB molecules are found to be 2.02 Å apart, less than the van der
16
Waals radius of 2.4 Å and close enough for dihydrogen bonds to form [37]. It is these hydrogen
bonds that give AB its high melting temperature and allow it to remain a solid at RTP unlike its
isoelectronic partner ethane. AB has a tetragonal unit cell at 300K, with cell parameters, a = b =
5.2630 Å and c =5.0505 Å [43], [44]. At 225K this shifts to an orthorhombic arrangement with
cell parameters a = 5.541, b = 4.705 and c = 5.0237Å [45].
Figure 8: The possible dihydrogen bonding arrangement in AB determined by Bowden et al. [44]. A
single layer on the ab plane is displayed. Nitrogen atoms are shown in pink and boron atoms are purple. a)
is the orthorhombic phase observed at 200K and b) is the tetragonal phase observed at 300K.
The bond distances in AB from various literature sources are collated in Table 6. These values
are gained from experiment and vary somewhat depending on the calculation method used [39].
Table 6: Bond distances in ammonia borane, theoretical and experimental sources.
Bond Bulk Material, Å Molecule, Å
Theoretical Neutron
Diffraction
Theoretical Microwave
spectrometry
[42] [46] [47] [43]
[42] [41]
N-B 1.59 1.59 1.58 1.58 1.65 1.67
B-HB 1.22, 1.23 1.22 ― 1.15, 1.18 1.22 1.21
N-HN 1.03 1.03 ― 1.07, 0.96 1.02 1.01
HN- HB
short
1.89 ― 1.91 2.02 ― ―
HN- HB
median
2.19 2.20 2.17 2.21 ― ―
HN- HB
long
2.22 ― 2.27 2.23 ― ―
17
Ammonia borane synthesis and regeneration
Ammonia borane can be synthesised in a variety of ways. The first published account by Shore
and Parry, which gives yields of 45%, is an exchange reaction with lithium borohydride [48],
𝐿𝑖𝐵𝐻4 +𝑁𝐻4𝐶𝑙 (𝑜𝑟(𝑁𝐻4)2𝑆𝑂4)
𝑑𝑖𝑒𝑡ℎ𝑦𝑙 𝑒𝑡ℎ𝑒𝑟→ 𝑁𝐻3𝐵𝐻3 + 𝐿𝑖𝐶𝑙 (𝑜𝑟 𝐿𝑖2𝑆𝑂4) + 𝐻2
( 5)
There is also a reaction involving base displacement: an adduct, 𝐻3𝐵 ∙ 𝑇𝐻𝐹, is formed by
dispersing diborane in tetrahydrofuran, THF; ammonia is added and the reaction takes place in
diethyl ether [49]
𝐻3𝐵 ∙ 𝑇𝐻𝐹 + 𝑁𝐻3
𝑑𝑖𝑒𝑡ℎ𝑦𝑙 𝑒𝑡ℎ𝑒𝑟→ 𝑁𝐻3𝐵𝐻3
( 6)
Higher yields of over 90% can be obtained with sodium borohydride and ammonium salts; an
example from Ramachandran and Gagare follows [50].
𝑁𝑎𝐵𝐻4 + (𝑁𝐻4)2𝐶𝑂3𝑇𝐻𝐹→ 𝑁𝐻3𝐵𝐻3 +𝐻2 + 𝑁𝑎𝐶𝑂3(𝑁𝐻4)
( 7)
The AB used in these experiments was sourced from Sigma Aldrich, 97% purity.
The hydrogen release from AB is an exothermic reaction process which leads to the formation of
many strong covalent bonds, making it difficult to dehydrogenate. However, regeneration routes
have recently been developed that remove this obstacle. Several methods exist [50], [51]; the one
detailed here enables regeneration from polyborazylene (NBH), the product when more than two
equivalents of hydrogen have been extracted using hydrazine [52].
Figure 9: AB regeneration with hydrazine and ammonia [52].
The reaction pathway of the dehydrogenation of ammonia borane
The hydrogen release from AB occurs in discrete steps, activated by temperature, during which
the 1st, 2nd and 3rd dihydrogen equivalents from each AB molecule are lost. The reaction is initiated
4 𝑁𝐻3𝐵𝐻3
4 𝑁2𝐻2 4 "𝑁𝐵𝐻"
5 𝑁2
10 𝑁𝐻3
−5 𝐻2
+15 𝐻2
10 𝐻2
18
in the material and propagates outwards from the nucleation site as a chain reaction. These steps
are sequential but overlapping; before the entire 1st equivalent of hydrogen has evolved, the
second step will have begun [53], [54]. Hydrogen evolution follows a sigmoidal form, indicative
of nucleation and then growth. The reaction equations for the release of the 1st, 2nd and 3rd
hydrogen equivalents are written below.
𝑛𝐵𝐻3𝑁𝐻3 → [𝐻2𝐵𝑁𝐻2]𝑛 + 𝑛𝐻2, <150°C [55] , <100°C [56], 110-130°C [57] ( 8)
[𝐻2𝐵𝑁𝐻2]𝑛 → [𝐻𝐵𝑁𝐻]𝑛 + 𝑛𝐻2 , 150-200°C [55], >100°C [56] ( 9)
[𝐻𝐵𝑁𝐻]𝑛 → 𝑛𝐵𝑁 + 𝑛𝐻2(𝑔) , > 200°C [55], >150°C [56] ( 10)
This general reaction scheme hides much complexity as the species can polymerise to form chains
or become terminated in rings [46], [58]. Chains are preferred if the temperature is kept low, while
rings are more probable under aggressive heating regimes [57].
Figure 10: Thermolysis of ammonia borane determined experimentally by Frueh et al.; both cyclic and
polymeric amido and imido boranes are produced [57] . Slow heating rates (<1°C min-1) do not produce
borazine, medium heating rates (3-10 °C min-1) show borazine during the release of the second hydrogen
equivalent and aggressive heating regimes (>10°C min-1) produce even more cyclic by-products [59].
The melting point of ammonia borane ranges in the literature from 104.5°C [60] 105°C [61],
through 110-112°C [62] and 112-114°C [36] to 112-117°C [55] and 121-124°C [63]. Most
sources agree that the melting peak is somewhat broad and begins at around 90°C, a few degrees
before the quoted value [60]. The melting is accompanied by the hydrogen release in step 1 which
starts simultaneously [55], [57], [62]. As soon as the AB molecules gain some conformational
freedom, the barrier to hydrogen release is significantly reduced and the reaction accelerates.
Ammonia borane
Amido boranes
Imido boranes
Borazine Amino borane
Boron nitride
19
The activation energy to hydrogen release from AB is high owing to the dihydrogen bonding
between molecules [54]. As with the melting temperature, a wide range of values for the strength
of these bonds are reported: 125kJ/mol [54] 150 kJ/mol [64] and 47.02kJ/mol [65].
It has been proposed that the first step can be broken down into three substeps: nucleation, where
AB softens and becomes mobile 𝐵𝐻3𝑁𝐻3∗; induction, where the diammoniate of diborane,
DADB, [𝑁𝐻2𝐵𝐻2𝑁𝐻2]−[𝐵𝐻4]
+, forms; and growth, where AB and DADB react together
releasing hydrogen [66].
Figure 11: The themolysis reaction mechanism of AB (polymeric) proposed by Stowe et al. showing the
induction, nucleation and growth phases [66].
One equivalent of hydrogen can also be released from AB while in the solid state – over several
hours – by maintaining the temperature between 70-90°C [53], [54]. However, at these
2
20
temperatures the barrier to the formation of DADB is high and so an induction period of several
hours precedes the hydrogen loss [56], [65], [67].
One of the more serious impediments to commercialising AB is the foaming that occurs during
dehydrogenation [65], [68], [69]. As the hydrogen release occurs after melting, large stable
bubbles form in the molten AB. The ensuing volume expansion can be in the range of several
thousand percent, see Chapter 4.
Modifying the dehydration properties of ammonia borane – literature study
Ammonia borane in water
AB dissolves in water (pH=9.1) [70] to about 25wt% [55]. AB at 25wt% can decompose in water
and release 9wt% H2 via this reaction pathway.
𝐵𝐻3𝑁𝐻3 + 2𝐻2𝑂
𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡→ 𝑁𝐻4
+ + 𝐵𝑂2− + 3𝐻2
( 11)
The AB-water solution is stable in an inert atmosphere but begins to release H2 if left in air. In air
the water becomes acidified by absorbing CO2 (to form carbonic acid) which catalyses the
hydrogen release [71]. Hydrolysis of AB is energetically favourable due to the formation of strong
B-O bonds and occurs at roughly 100 times the rate of AB thermolysis [55]. The rate of hydrogen
release only reaches acceptable levels if catalysts such as acids or transition metals are used [72]–
[76].
The hydrolysis of AB has several advantages over the thermolysis of AB. Once dissolved in water
the AB can be flowed making the material easier to handle and the dehydrogenation easier to
control. Additionally the AB reaction products are insoluble in water and therefore can be
separated from the still unreacted AB. However, due to the weight of water, hydrolysis of AB is
unlikely to meet the DOE system weight requirements for on-board hydrogen storage.
Ammonia borane in alternative solvents
AB is soluble in various other polar solvents, many of which are shown to improve the AB
reaction kinetics and thermodynamics. Ethers such as glyme, diglyme, tetraglyme, THF and 2-
methyl THF have been particularly well studied [77]–[80]. In ethereal solvents AB decomposes
via the two-step mechanism. However, instead of producing polymeric PAB type species the
hydrogen release leads to cyclic products such as cyclotriborazane and borazine [67], [79]. In fact,
AB in diglyme (at 130°C) and tetraglyme is considered a suitable method for producing borazine
with hydrogen as a side product [77].
21
3𝐵𝐻3𝑁𝐻3
tetraglyme,150°C,3hours→ 𝑁3𝐵3𝐻6 + 6𝐻2
( 12)
AB reacting in the presence of these solvents releases hydrogen at accelerated rates at lower
temperatures. The lack of induction period has been linked to the early appearance of DADB,
observed in NMR studies [78], [80]. Colquhoun et al. showed that hydrogen bonds formed
between the HNδ+ on the AB and the electron-donating oxygen on the ether [80], [81]. AB has
been observed to react in acetonitrile to produce hydrogen and ethylamine-borane, C2H5-NH2BH3.
[79].
Figure 12: A) Decomposition pathway for AB in glyme suggested by Shaw et al. based on boron NMR
experiments [78]. AB follows that cyclic reaction route. B) From Kim et al. showing how hydrogen
bonding between AB and ether promotes formation of DADB [80].
Ammonia borane in ionic liquids
Ionic liquids have also been shown to have an accelerating effect on AB dehydrogenation while
limiting the production of borazine [67], [82]. Wright et al. have written an excellent review on
metal catalysed AB dehydrogenation in ionic liquids. Yields of greater than 2.2 equivalents of
hydrogen can be achieved at increased rates compared to pristine AB [83]. The AB follows the
polymeric reaction route and so the release of borazine is limited.
A B
22
Ammonia borane in acidic conditions
Acidic conditions cause rapid hydrogen release from AB [48], [62]. The rate determining step in
the reaction is the B-N bond cleavage [37], [84]. This is achieved by attack with the electrophilic
H+ which leads to NH4+.
Brønsted and Lewis acids were shown to encourage the breaking of the B-N bond [37]. Hydrogen
release was enhanced but cyclic products were also generated.
Figure 13: The mechanism of acid dehydrogenation of AB proposed Stephens et al. [37]. The acid
abstracts a hydride (a) activating the boron to interact with a neighbouring AB molecule (b) forming an
intermediate. Hydrogen loss follows and an intermediate with a bridging hydrogen is formed. If the acid
to AB ratio is low dehydropolymerization ensues (c). However, if the ratio is high, a second reaction
takes place (d) which catalyses the reaction further (e).
An interesting study that combined both catalysis and nanostructuring was conducted by Zhang
et al. AB (20wt% only) was combined with silica nanospheres (coated with acidic hydroxyl
groups) [85]. Earlier hydrogen release with supressed borazine and ammonia was observed. A
similar effect was observed by Stephens et al. with their study on AB in hydroxyl-coated carbon
cryogel [84]. Acid additives have a good tradition of encouraging improved dehydrogenation
characteristics in AB.
23
Zeolites and metal organic frameworks
Zeolites are highly porous, low density, crystalline aluminosilicates, cheaply available and simple
to synthesise. Zeolites provide an ideal hosting environment for various metal nanoclusters that
can be used to catalyse AB. Examples include Pd, Rh, Ni and Co [65], [74]. While zeolites
favourably improve the AB hydrolysis, often the overall the system weight is too high to meet the
DOE targets [11].
Zeolite-X (containing caesium active sites) and Chabazite (casium substituted potassium) were
ball-milled with AB, (ratio AB:zeolite 10:1) [86]. At low heating the zeolites reduced the
induction period significantly and at high temperature they increased the amount of hydrogen
released. Borazine levels were reduced but not eliminated.
Carbon
Carbon by itself is not active in the catalysis of ammonia borane. However activated or
functionalised carbon can provide a combined nanoscaffold and catalysis environment [64], [84],
[87].
AB loaded on a carbon cryogel (CC) released hydrogen faster than pristine AB at 85°C [87]. The
increased dehydrogenation rate was attributed to defect sites, destabilisation of the AB hydrogen
bonding network and the AB reacting with the hydroxyl coated surface (B-O bonds were formed).
Borazine release was suppressed. The pore size of these AB-CC composites was found to be
directly correlated to the temperature of the hydrogen release [64], [84]. The greater the surface
area to volume ratio the faster the dehydrogenation.
Theoretical DFT studies have suggested the internal spacing of a carbon nanotube could provide
the ideal environment to ameliorate the dehydrogenation of AB [88].
AB (30wt%) loaded on the graphene oxide released hydrogen at lower temperatures than pristine
AB; additionally the complete suppression of impurities including borazine, ammonia and
diborane was observed up to 200°C [89]. This material is particularly interesting as the AB could
be regenerated in situ with hydrazine and ammonia, 75% and 61% of the original AB was
recovered in the first and second cycles.
Nanostructured AB (50wt%) was incorporated into a mesoporous carbon framework studded with
Li crystallites (5wt%) [90]. Hydrogen release was accelerated such that significant amounts could
be achieved at 60 due to the confinement and catalysis due to Li. The gaseous impurities where
also controlled.
24
Metal catalysts and ammonia borane
AB was combined with CoCl2, NiCl2 and CuCl2 by mechanical mixing (metal content 15wt%).
The Cu2+ ion was the most effective catalyst causing 2 equivalents of hydrogen to be released at
60°C without incubation time [91], [92]. Decomposition was shown to proceed via an alternative
reaction route which includes the intermediate[𝑁𝐻4]+[𝐵𝐶𝑙4]
−:
𝑁𝐻3𝐵𝐻3 + 2𝐶𝑢𝐶𝑙2⟶ [𝑁𝐻4]+[𝐵𝐶𝑙4]
− + 𝐻2 + 2𝐶𝑢(0) ( 13)
This route does not include DADB and this is part of the reason no borazine is released.
CoCl2 or NiCl2 (2mol %) were coprecipitated with AB. The hydrogen release from this material
was observed as low as 59°C. No borazine was released and there was no perceptible foaming
[92].
Alkali metal hydrides have also been used to destabilise AB and improve its reaction kinetics.
LiH and NaH were ball milled with AB to produce LiNH2BH3 and NaNH2BH3 [93], [94]. The
complexes demonstrated improved reaction kinetics at low temperatures (<100°C), reduced
foaming and limited borazine release. In other another study, AB was combined with NaMgH3 by
ball-milling in a ratio of 3 AB to 1 NaMgH3. 10wt% H2 was released from the mix in 2 minutes
at 80°C, and the borazine presence in the released gas was reduced [95]. Recently, Mg(NH2BH3)2
has been synthesised by ball-milling Mg and AB together. The resulting material could release
~10 wt% high purity hydrogen below 300°C [96].
Clearly metals are a highly efficient way of catalysing AB.
Ammonia borane in the presence of polymers
Poly(vinyl pyrrolidone) (PVP) was combined with AB via single phase electrospinning [97]. With
a 20wt% AB content the hydrogen was released at lower temperatures as compared to pristine
AB but the ammonia levels were increased. This necessitated the addition of MgCl2 to mop up
the released ammonia.
Polyacrylamide (PAM) with AB in a 1:1 ratio demonstrated a significant improvement in
hydrogen release kinetics and thermodynamics. However, ZnCl2 was required, as excess ammonia
was evolved [98]. It was suggested that AB was interacting with the carbonyl group on the PAM
at the AB-PAM interface to release hydrogen contaminated with ammonia, but that also bulk AB
was releasing hydrogen in the traditional manner.
A similar effect was observed for AB confined with poly(methyl acrylate) (PMA) with AB:PMA
ratios 8:10 and 2:10 [99]. In both cases hydrogen was released earlier than in the case of pristine
25
AB. The borazine level in the gas stream was reduced but the ammonia level increased. Boron to
oxygen bonds were observed and found to be linked to the increased ammonia level.
Ammonia borane was successfully combined in polystyrene (PS) nanocavities by co-axial
electrospinning. The dehydrogenation temperature of the AB was significantly reduced compared
to pristine AB and some impurities were filtered out by the PS [100]. However, the foaming could
not be controlled and the AB mass content was only around 30wt%.
Materials used and project aims
Polymers
2.3.1.1 Polyethylene oxide
The hydrogen release of AB is preceded by melting and conversion to DADB. The melting
increases the conformational freedom of the molecule, reducing the energy barrier to
dehydrogenation. The melting point of AB is 112-114°C [55], [58], but melting typically begins
a few degrees lower. While AB can release hydrogen at low temperatures (<100°C) while in the
solid phase [53], the process is slow and several hours are required to release 1 equiv. of H2. If
the melting point of AB could be lowered, rapid hydrogen release could occur at lower
temperatures.
Polyethylene oxide (PEO) is a polymer with a low melting point (66-75°C) available in a large
range of molecular weights, 10 - 10,000,000 (Figure 14: Schematic of the PEO monomer
unit.Figure 14). PEO is cheap and non-toxic and is often used as a binder for medicines. The
shorter chains are termed polyethylene glycol (PEG) and can occur in a liquid state as the melting
point decreases with decreasing chain length [101]; the monomer, ethylene glycol C2H4(OH)2, is
a liquid under standard conditions. PEO is soluble in mostly polar solvents including water,
acetonitrile, benzene, alcohols, chloroform, esters, cyclohexanone and N,N-dimethylacetamide
[101].
Figure 14: Schematic of the PEO monomer unit.
The PEO chains twist into helixes which can be arranged in a variety of crystal arrangements, the
most thermodynamically stable of which is monoclinic with a=8.05, b=13.04, c=19.48Å and β=
125.4°.
26
A composite can defined as a homogenous material, composed of two or more heterogeneous
substances where the final mix demonstrates superior properties to either of the constituent parts
[102]. For example, the combination of calcium carbonate and collagen in bone provide both
strength and flexibility and adding glass fibres to polystyrene gives it more load-bearing
capability. The properties of polymer composites are often hard to predict; the polymer types and
their interactions, the proportions, the mixing process and the level of crystallinity are some of
the many parameters to consider. However, it is still possible to tailor the resulting properties of
the blend to provide the flexibility, radiation resistance, transparency, melting point etc. that is
required. PEO has a melting point of 66-75°C [103]. If correctly mixed with AB it could bring
the melting point of AB down and activate the hydrogen release at lower temperatures.
Nanostructuring AB has been shown to aid dehydrogenation; graphite [89], silicone glass [87]
and polystyrene [100] scaffolds have all proved helpful. Electrospinning could be an ideal method
to blend PEO and AB at the nano level making not only a composite with low melting point but
an AB nanostructure as well.
The thermodynamic properties and structure of AB are defined by its ability to form dihydrogen
and hydrogen bonds. The repeat unit of PEO is depicted in Figure 14; the lone pair of electrons
on the oxygen is capable of forming hydrogen bonds. The polar nature of both AB and PEO will
facilitate their intimate mixing via electrospinning. We expect also to form hydrogen bonds
between the AB and PEO.
2.3.1.2 Polystyrene
The dehydrogenation kinetics and thermodynamics of AB can be favourably modified not only
by chemical catalysis [104] but by nanostructuring [84], [100], [105]. As particles get smaller, the
diffusion distances shorten and this can lead to faster reaction kinetics. Also, surface energy
changes can arise which alter the thermodynamic stability of the material enabling the hydrogen
release to follow a different, lower energy, reaction route [106].
In particular, nanostructuring AB in coaxially electrospun polystyrene (PS) fibres has reduced the
reaction temperature 15-20ºC [100]. Kurban et al. tuned the miscibility, conductivity and
viscosity of the core (internal, containing AB) and shell (external, containing PS) solutions to
allow mixing during spinning to produce highly porous fibres. The encapsulated AB
demonstrated properties suggesting a nanostructured state. However, the maximum AB content
achieved in the fibres that showed a significant reduction in the hydrogen release temperature was
only 30wt%. Furthermore, the mass loss during reaction indicated that AB sublimed and foamed.
In chapter 4, the reaction kinetics and thermodynamics of AB are improved by combining with
PEO. Additionally it suppresses foaming, but it encourages impurities such as borazine. An
27
improvement on this system should further limit foaming and also cut down the impurities. Being
unable to accurately determine the particle size of AB in the AB-PEO composites, it is possible
that the diminished activation energy is at least partly due to nanostructuring in the AB. It would
therefore be useful and interesting to isolate and investigate the behaviour of nanosize particles
of AB in a polymer.
Figure 15: Schematic of the PS monomer unit.
PS is ideal polymer to use as a nanoscaffold. It has a high glass transition temperature, 100 ºC,
and melting point, 240 ºC, so will remain solid as the AB reacts. There are no active functional
groups so the effect should be purely physical. PS is a very well-understood material and has been
electrospun extensively in the past [107]. In order to build on the previous work of Kurban et al.
I will use single phase electrospinning. Single phase electrospinning is simpler then coaxial as
there are fewer parameters to regulate and should allow a higher amount of AB to be incorporated
in the fibres. To this end I will require a solvent that can dissolve both polar (AB) and nonpolar
(PS). I wish to employ the repulsion between the polar and nonpolar entities to divide the AB and
PS into discrete, hopefully nanostructured, areas.
This part of my thesis considers the unusual practice of electrospinning polar and nonpolar
materials in a single phase. The spinning parameters necessary to produce AB-PS fibres will be
explored and the material properties − hydrogen release temperature, impurity content in the gas
stream, and foaming − can be studied and compared to pure AB.
Clays
The hydrogen release of AB has been promoted by being combined with PEO. The temperature
of the onset of hydrogen liberation has been reduced and the foaming and expansion limited by
the PEO. However, the proportion of the impurity borazine has increased in the gas stream. These
effects are due in part to the physical behaviour of the polymer and in part to the nature of the
functional group, the ethyl group.
Previous work has shown that an alternative functional group, the hydroxyl, can positively
influence the dehydrogenation of AB. Materials tested include mannitol [108], hydroxyl-coated
graphene oxide [89], carbon cryogel [87] and silica nanospheres [85]. In all cases the temperature
28
for the onset of hydrogen release was reduced and borazine was eradicated from the gas stream.
These studies were particularly successful as in all cases the AB was also nanoconfined,
something that was not achieved in the PEO-AB.
Clays are layered hydrous aluminosilicates covered with hydroxyls [109]. Sheets of tetragonally-
coordinated silicon are layered with octahedrally-coordinated aluminium terminated with
hydrogens that give the surface its acidic nature. The layers are separated by adsorbed water and
the layers can slide over each other giving clay its plastic, malleable character. Variety is achieved
as the aluminium can be substituted by iron and magnesium. Alkali and alkaline-earth metal
cations are also found in the interlayer space. Clays are cheap, abundantly available, non-toxic
and environmentally benign.
If AB was confined in the interlayer space, nano-structured AB could be attained in close
proximity to catalytic hydroxyl groups.
2.3.2.1 Smectite clays, laponite and montmorillonite
Smectite clays are layered and each layer consists of two sheets of octahedrally-coordinated
aluminium oxide on either side of a tetragonal silicate sheet [109]. Substitutions in the Al and Si
sites create charge imbalances which are stabilised by Na and Ca counterions between the layers.
Smectites are swelling clays and the layers are expandable by water, which is pulled in to hydrate
the counterions. The interlayer separation means that the layer stacking is fairly disordered.
Heating dehydrates the interlayer space, but the process is reversible and the clay can be
rehydrated. Other polar liquids and salts can be intercalated into the layers.
Two smectites were tested: montmorillonite, a natural clay found in sedimentary rocks, and
laponite [110], a synthetic smectite clay where the octahedrally-coordinated silicon has been
completely replaced by magnesium. The counterion in the montmorillonite used was sodium and,
in laponite, sodium and a small amount of lithium are present. In laponite the layers huddle
together to form nanometre-sized disks, about 25 times wider than they are high. When hydrated,
the counter ions are liberated which results in negatively-charged OH- groups on the disk surface,
while at the rim positive charges congregate. Laponite dissolves readily in water and forms gels
at low concentration with a pH level of 10.
The empirical formula of montmorillonite is Na0.3Al2Si4O10(OH)2(H2O)10 and for laponite it is
Na+0.7[(Si8Mg5.5Li0.3)O8(OH)4]-0.7.
Smectites could improve the dehydrogenation properties of AB. AB in the interlayer spacing
would be nano-confined and within reach of catalytically active hydroxyl groups.
29
Sodium montmorillonite SWy-1 from Wyoming, Crook County, USA and laponite RD sourced
from Rockwood Additives Limited were both used as received.
2.3.2.2 Tubular clays, imogolite and halloysite
Imogolite is a tubular clay of the general formula (OH)3Al2O3SiOH [111] found in volcanic ash
[111], [112]. It is formed of a gibbsite Al(OH)3 sheet coupled on one side to a tetrahedrally-
coordinated silicon oxide layer [113]. The Si−O bonds are shorter than Al−O bonds, the sheet
curls into a tube displaying the gibbsite (AlOH) on the external surface and the siloxane (SiOH)
on the inside. Both surfaces are positively charged, the intra-wall oxygen carries the balancing
negative charge [113]. The hydroxyl groups are both acidic but as the Si-OH bonds are more ionic
then the corresponding Al-OH bonds and should therefore be more active.
The imogolite used is synthetic, sourced industrially from Kodak. The tubes are approximately
100nm in length with an external diameter of around 20Å and an internal diameter of 10Å [114].
The tubes collect into bundles, with a repeat centre to centre separation distance of 23Å for natural
imogolite [114] or 28Å in the synthetic case [111].
Hydrophilic behaviour occurs on both internal and external surfaces [113]. Water uptake occurs
inside the tubes and in the vacant spaces between tubes [113]. As the water level increases the
individual tubes become solvated and the bundles separate. The tubes gain motional freedom and
can loss their circular diameter becoming elliptical [111]. Imogolite water mixes have a gel phase
at low imogolite concentration.
Imogolite is both biocompatible and environmentally benign [115]. The small size of the tubes
provides a very high surface area per unit mass. Through freeze drying with AB it should be
possible to exploit the imogolite active external surface to catalyse hydrogen release from AB.
The hydrated ammonia borane will likely be too large to enter the tubes with the water molecules,
so the AB will only be exposed to the gibbsite external surface. It would be interesting to study
also the effect of the siloxane surface on AB and to make a comparison to the gibbsite. For this
halloysite clay nanotubes would be ideal.
Halloysite is a naturally occurring kaolinite clay found around the world in various soils but
particularly in volcanic areas [109]. A layer of SiO4 tetrahedra is bonded to a gibbsite like layer,
out of every three sites, only two are filled by Al to give a final composition of Al2Si2O5(OH)4.
The layers are separated by a single disordered layer of water to a distance of 10Å. While it can
take many shapes, the Al vacancies cause the layers to curl up into spiral tubes to form the most
common morphology, multi-walled tubes. Contrary to the imogolite, the halloysite tubes keep the
siloxanes on the external surface leaving the gibbsite layer on the interior of the tube.
30
The internal layer spacing in halloysite is suitable for the intercalation of salts and organic
compounds. Complexes can be achieved if the intercalant is polar [116]. Aqueous soluble salts
can be intercalated into hydrated halloysite to form monolayers between the layers. If the water
has been dried off, the interlayer spacing will not be amenable to rehydration or other liquids. At
pH 6-7, the halloysite SiO2 surface has a negative charge while the Al2O3 interior surface is
positive [117].
It should be possible to load AB into and around the halloysite nanotubes using water as a carrier.
Freeze drying the mixture will create an AB halloysite composite. The halloysite might improve
the dehydrogenation properties of AB either as a nanoscaffold or through the catalytic nature of
the hydroxyl groups.
Halloysite nanoclay was obtained from Sigma Aldrich and used as received.
Graphite intercalation compounds
Here a study of the motion of ammonia inside a calcium ammonia (Ca-NH3) graphite intercalation
compound (GIC) is presented. Calcium and ammonia are highly compatible; calcium readily
dissolves in ammonia to make a highly-structured octahedrally-coordinated 3-dimensional liquid,
Ca(NH3)6. In a GIC, however, the confinement in graphite restricts the vertical dimension and
only allows a maximum of four ammonia to approach each metal centre [26], [34], [118], [119].
In this way the 3-dimensional solution is constrained into a 2-dimensional analogue. Here the
motions of the ammonia in a Ca GIC will be studied. The motions of confined ammonia can then
be compared to previous studies of the unhindered liquid Ca(NH3)6 and other metal ammonia
GICs [120].
Previous studies on ammonia in similar materials have shown rotational modes at low
temperatures which develop into diffusion on heating. In Ca(NH3)6 [120] and in KC24(NH3)4.3,
the potassium analogue of the material here studied, [121]–[123] the onset of movement at ~50K
is the rotation of the hydrogen atoms about the fixed metal to nitrogen axis. Above between 100
and 200K a second rotational mode occurs in both complexes: in Ca(NH3)6 the whole group is
slowly reoriented about the centre of mass [120], while in KC24(NH3)4.3 the ammonia molecules
wheel around the potassium centre [121]–[123]. High temperature studies of Ca(NH3)6 have not
been published but for KC24(NH3)4.3 raising the temperature to 300K increases the energy of the
rotation and translational diffusion at 10-5cm2s-1 becomes apparent.
The final chapter of this thesis reports on studies of the translational and rotational motions of Ca-
NH3-GIC; a temperature dependent study has not at the time of writing been performed on this
material. Time of flight neutron spectroscopy was used to follow the changes in the behaviour of
31
the ammonia at 50K intervals between 300K and 2K. The large incoherent scattering cross section
of the proton was employed to follow the movement of the ammonia as it dominates the measured
signal. We show that the diffusion at the higher temperature switches to a rotational mode at 200K
which fades below 150K. Further we show that NH3 cannot be completely deintercalated.
32
3 Experimental methods
Making samples
Electrospinning
Electrospinning is a powerful technique for the production of micro and nano sized material
structures, usually long filaments, beads or beaded fibres [107]. The method is a highly versatile
and controllable; the composition, shape and architecture are down to the inventiveness and skill
of the technician. A one-step process, electrospinning is practically scalable for high volume
production of fibres with a wide range of industrial uses. Nonwoven fibrous meshes suitable for
wound dressing, air filtration or composite reinforcement have all been produced. More complex
arrangements have encapsulated drugs, anti-cancer and antibiotics, into the fibres for controlled
release [124], or carbon nanoparticles for battery anodes [125], and also hollow microspheres for
containing proteins [126].
The base component of the solution used in electrospinning is polymeric; the polymers provide
the necessary viscosity and entanglement to produce long continuous fibres [107]. In a traditional
electrospinning setup, the solution is passed at a constant rate of flow, say 1 mlhr-1, through a
suspended hollow needle attached to a voltage supply. A droplet appears at the nozzle and
becomes infused with charge; this potential can eventually overcome the surface tension and
provoke a jet that moves towards to an earthed, or oppositely charged collection plate. As the jet
extends, instabilities cause it to bend and whip around in a spiral motion, this serves to stretch and
lengthen the polymer chains within the strand as it descends. The fibres or beads pile up and stick
together to form a mat. The solution composition defines the integrity of the jet, and the
morphology, beaded or fibrous, of the final product.
Non polymeric compounds can also be processed via electrospinning either solely or incorporated
with polymers. This is useful for developing composites with unique properties and functionality
such as fluorescence, conductivity or increased heat resistance. The simplest method is to make
and electrospin a single solution containing all the necessary ingredients, however the final fibre
will be limited to a single mixed state and a co-solvent for all the additives may not be available.
To create more interesting architectures concentric needles for distinct solutions can be used.
Coaxial electrospinning [107][100] allows two initially separate solutions to interact during the
descent and produce a variety of structures. Electrospinning was found to be a most suitable
technique to make closely mixed AB polymer composites.
33
Figure 16: Single phase electrospinning rig.
There are both solution and apparatus parameters that need to be optimised to achieve the desired
product. The first step is to choose the polymer and to dissolve it in a suitable solvent. The solution
must be viscous and conducting. The flow rate must be set and the voltage and needle to collector
distance adjusted to ensure the fibres have time to dry before they are laid down on the collector.
3.1.1.1 Solution parameters
A polymer is a chain of covalently linked, repeating monomer units, it is characterised by the
monomer chemistry and the chain length, 𝑀𝑖. As the chain length increases the polymer glass
transition and melting point will rise and it will become harder to dissolve [103], [127]. In a
polymer sample there will always be a distribution of molecular weights; so usually an average
Hollow needle
High voltage source
Earth
Whipping strand
Droplet
Collection plate
Nozzle
Solution
34
value is quoted, either the number average molecular weight, 𝑀𝑛, or the weight average molecular
weight, 𝑀𝑤.
𝑀𝑛 =
∑𝑁𝑖𝑀𝑖
∑𝑁𝑖 and 𝑀𝑤 =
∑𝑁𝑖𝑀𝑖2
∑𝑁𝑖𝑀𝑖
( 14)
where 𝑁𝑖 is the number of chains of molecular weight 𝑀𝑖. The polydispersity of a polymer sample,
𝑃 = 𝑀𝑛 𝑀𝑤⁄ , is influenced by the polymerisation method; free radical polymerisation will give a
large 𝑃 while condensation polymerisation gives a small P [128]. For electrospinning fibres, long
polymer chains are desired to maintain the integrity of the jet on the other hand, making beads
requires short chain polymers that allow the jet to segment [129]–[131]. Low polydispersity is
preferred in both cases as it increases the quality and homogeneousness of the fibres or beads
[132].
3.1.1.2 Solvent
A polymer dissolves in stages; the solid mass softens and swells before the chains peel off into
the solution [127]. Areas of crystallinity are tough to dissolve as the tight packing of the polymer
chains makes it difficult for the solvent to permeate. Dissolving is energetically favourable
because the entropy or disorder, ∆S, of the system increases. Raising the temperature will make
dissolving more energy favourable as quantified via the Gibbs equation for free energy,
∆𝐺 = ∆𝐻 − 𝑇∆𝑆 ( 15)
where, ∆𝐺 is the Gibbs free energy and ∆𝐻 is the enthalpy of the mixing. The chance of forming
a mixed solution increases as ∆𝐺 becomes more negative i.e. the smaller the change in ∆𝐻 the
more likely the polymer is to dissolve. ∆𝐻 is determined from the solubility parameters, δ, and
partial volume fractions, ν, of the constituent polymer and solvent, respectively subscripts S and
P.
∆𝐻 = 𝑣𝑆𝑣𝑃(𝛿𝑆 − 𝛿𝑃)2
( 16)
The smallest ∆𝐻 is obtained when the solubility parameters of the polymer and solvent are most
similar, in other words, like dissolves like [127].
The solubility parameters are actually a description of the disruption of the cohesive energy;
molecules in a material are tied together by dint of their cohesive energy, 𝐸𝑐𝑜ℎ. The cohesive
energy itself can be broken down to its contributions of the polar, dispersive and hydrogen
bonding energies in the material, 𝐸𝑝 , 𝐸𝑑 and 𝐸ℎ respectively
35
𝛿 = √𝐸𝑐𝑜ℎ𝑣𝑚
= √𝐸𝑝 + 𝐸𝑑 + 𝐸ℎ
𝑣𝑚
( 17)
where, 𝑣𝑚, is the volume of the material. The solubility parameter can also be deconvoluted into
its polar, 𝛿𝑝, dispersive, 𝛿𝑑, and hydrogen bonding, 𝛿ℎ, parts.
𝛿2 = 𝛿𝑑2 + 𝛿𝑝
2 + 𝛿ℎ2
( 18)
Like dissolves like can now be practically exploited; for example high polarity solvents are
suitable for high polarity polymers while non-polar solvents dissolve non-polar polymers.
In addition to being a good solvent for the polymer in question, the solvent must carry the charge,
vaporise rapidly to ensure the fibres are dry at the collection plate and provide surface tension.
3.1.1.3 Electrical conductivity
Conductivity, C, is a measure of the current density, 𝐽, that can be induced by an electric field, E.
𝐶 =𝐽
𝐸=𝑝𝑣𝑐𝐸
( 19)
where p is the charge per unit volume and v is the drift velocity of the charge. Electrospinning is
dependent on the conductivity in the solution which carries the charge, which overcomes the
surface tension, to the extremities of the droplet. The current is carried by ions or whole molecules
as the electrons in liquids are usually confined to their respective molecules.
As the conductivity or dielectric constant of the solution increases it becomes more sensitive to
the electric field [133]. Electrospinning will begin at lower voltages and instabilities on the droplet
leading to multiple jets are more likely to occur [134], [135]. With more effective stretching forces
on the jet, the incidence of beading will decrease [135], and the fibres will become finer and more
uniform [131], [136]. A minimum amount of conducting species are necessary in the mixture
[107] as solutions with zero or low conductivity will pour or drop from the nozzle without
spinning [135]. The dielectric response (or permittivity), ε, of the solution to the field must also
be considered as static charge build up can initiate an electrospinning jet. An electric field incites
a polarity, P, in the molecules, which further orientate themselves to oppose the field. The level
of reaction to the field is determined,
𝐸 = 휀𝑜𝐸𝑙 + 𝑃 = 휀𝑜휀𝐸𝑙 ( 20)
where the field, εo, the permittivity of free space, and the resulting field within the liquid, E.
36
Of the solvents used in electrospinning, most have very low conductivity (10-3 to 10-9 ohm m-1) as
there are very few ions available to carry charge [107]. Additives can be used to increase the
conductivity of the solution and these include solvents such as DMF or water [137], or non-
interacting species such as mineral salts, mineral acids, carboxylic acids or tetraalkylammoniaum
salts [107]. The conductivity of the solution is directly proportional to the amount of salt added
[135]. The type (size) of the charge carries affects how easily the charge can be distributed in the
electrospinning process and this affects the produced fibre, smaller diameters resulting from the
addition of higher density charge carriers [138]. The affect is also noted in coaxial spinning set-
ups [136].
Significantly, the applied charge can be positive, negative or alternating [107]. A negative charge
has been shown to give a lower diameter distribution in the fibres as it is easier to dissipate
negative electrons then positive ions.
3.1.1.4 Volatility
It is important to control and balance the evaporation of the solvent during electrospinning [107].
The evaporation should be fast to prevent the deposition of wet fibres, but excessive drying should
be avoided as this can lead to solidification at the nozzle and tube blockage. The fibre diameter
can be controlled by the volatility of the solvent; the longer the whipping jet remains fluid, the
further it can stretch and the thinner the resultant fibres will be. The rate of vaporisation from the
jet is down to the vapour pressure of the solvent. The vapour pressure, 𝑃𝑣𝑎𝑝, can be determined
from the Clausius-Clapeyron equation and is related directly boiling point, 𝑇𝐵, and heat of
vaporisation, ∆𝐻𝑣𝑎𝑝, of the solvent.
The vapour pressure increases with the actual temperature, 𝑇𝑜, and decreases with a lower
atmospheric pressure, 𝑃𝑎𝑡𝑚.
𝑃𝑣𝑎𝑝 = 𝑒𝑥𝑝 [−∆𝐻𝑣𝑎𝑝𝑅
(1
𝑇𝐵−1
𝑇𝑜)]𝑃𝑎𝑡𝑚
( 21)
37
Table 7: Solvents used in practical work, boiling point and vapour pressure.
Solvent Boiling point, °C Vapour pressure mmHg at 20 °C
Water 100 17.5
Ethanol 78.37 44.6
Acetonitrile 81-82 72.8
Dimethyl sulphoxide 189 0.42
Tetrahydrofuran 65-67 143
Dimethyl formamide 153 2.7
If moist fibres are being produced, raising the temperature around the jet can accelerate drying.
Conversely, if excess drying of the droplet is occurring, a gas jacket containing the spinning
solvent, can be used around the nozzle to suppress the rate of evaporation [139], [140]. Another
feature of a using a highly volatile solvent in electrospinning is the fabrication of flat fibres; when
the solvent near the surface evaporates and a hard crust forms. This later collapses into the space
vacated as the interior liquid evaporates.
3.1.1.5 Viscosity
The viscosity, 𝜂, of a solution is the strain rate, �̇�, under a particular stress, 𝜏; 𝜂 = 𝜏 �̇�⁄ [141],
[142]. Polymer solutions are naturally viscous as the entangled chains are difficult to displace. A
viscous polymer solution is best achieved with a high concentration of long, branched, well
solvated and unfurled polymer chains.
The polymer concentration affects the viscosity according to this power law,
𝜂 ∝ 𝐶𝜂𝛽 ( 22)
where 𝐶𝜂 is the polymer concentration and β is a constant dependent on the solvent polymer mix
[143] and the molecular weight, 𝑀𝑤 , and the effective volume of the polymer chain, (𝑟2)3 2⁄
influence the viscosity.
𝜂 =
𝐾(𝑟2)3 2⁄
𝑀𝑤
( 23)
where K is the proportionality constant from the Flory equation for hydrodynamic volume and r
is the polymer end to end distance [129].
Smooth unbroken fibres can only be achieved if the polymers in the solution can entangle during
the spinning process [131], a minimum concentration Ce is defined as the minimum polymer
concentration necessary to move away from a bead regime [144] and form connected fibres.
Upper and lower limits on the allowed polymer concentration are determined by the chain length
38
and the interactions between the specific polymer and solvent chosen. As the chain length of the
polymer and the level of branching increases, Ce may be achieved at lower concentrations.
The fibre diameter, d, is related to viscosity, 𝑑 ∝ 𝜂𝛼 where α is a scaling component different for
every solution. A thicker solution will create larger diameter fibres, but this is also limited as if
the viscosity is raised too high the solution will become too thick to pass easily through the nozzle
[107]. The higher the viscosity the higher the voltage needed to produce a jet at the nozzle and
the smaller the deposition will become as the viscosity can counteract the bending instability
forces on the jet. This will lead to fatter fibres.
Temperature affects viscosity by increasing the solubility of the polymer,
[𝜂] = 𝐵𝑒𝐸𝑎 𝑅𝑇⁄ ( 24)
where, B is a constant, Ea is the activation energy for viscous flow, R is the gas constant and T
the temperature. Usually increasing the temperature will reduce the viscosity by allowing the
chains to reptate, but if the solvent is poor than at higher temperatures the polymer chain will
unfurl and increase the intrinsic viscosity [107].
Polymer solutions undergo viscosity changes due to shear thinning. Shear thinning is significant
in highly concentrated solutions such as those used in electrospinning. In the natural state the
polymer chains are entangled. The viscosity changes under elongation depend on the rate of
change of the solution stretch. At small strain rates the polymers can reptate (wiggle) under
thermal fluctuations of the polymer chain and the system remains close to equilibrium. At high
strain rates, the tubes (free volume the polymer migrates in) are aligned in the direction of the
strain and the viscosity is reduced. At high strain rates the polymer chains inside the tubes are
forced to stretch as the tubes become thinner and the viscosity increases again [142].
3.1.1.6 Surface tension
Surface tension arises due to an imbalance of forces on the molecules at the liquid vapour barrier
on a liquid surface [129], [145]. The surface tension of the solution works to reduce the unit area
per unit mass, in practice it causes water droplets to pull into a ball. If we want to alter the shape
of a droplet away from its preferred spherical form we will be increasing the surface area and
hence the energy of the liquid.
𝜉 = 0.3
𝐿
𝑁𝐴(𝑁𝐴𝜌
𝑀𝑤)2 3⁄
( 25)
where the surface tension, 𝜉 is related to the latent heat of vaporisation, L per mole, NA , the
molecular weight, 𝑀𝑤 and the density, ρ.
39
In a polymer solution the surface tension is affected by the solvent type [137], [146], the polymer
concentration, chain length, conformation and polymer solvent interactions [130], [147]. The
stronger the intermolecular forces in the solvent the higher the surface tension; water with its
hydrogen bonding has a higher surface tension than hexane that is governed only by dispersive
forces [145]. The polymer end to end length, not only controls the viscosity but the surface tension
as well; an elongated polymer with good solvent-to-polymer interactions will vastly increase the
surface tension.
During electrospinning surface tension works to reduce the surface area while the stretching
forces work to increase the surface area of the spinning jet [129]. The surface tension of the
solution is closely related to the critical voltage of electrospinning, the minimum voltage that can
force a jet from the droplet [137]. A high surface tension is useful when fabricating beads as it
promotes faster jet division and moulds the fragments into spheres [126].
When electrospinning, it may often be desired to change the surface tension of the solution to
ameliorate the fibre morphology. Utilising a different solvent or a solvent mix has been shown to
affect the surface tension by more than 10% [137]. Varying the polymer concentration up or down
will modify surface tension; switching the polymer type is also an option. Surfactants, alcohols
and organic molecules with a polar end, may be added to a water-type liquid to reduce the surface
tension while inorganic salts can strengthen the intermolecular forces [145].
Spinning parameters
3.1.2.1 Voltage
The voltage provides the surface charge on the jet and the electric field between the needle tip
and the collector plate. Charge builds up in the droplet until it can overcome the surface tension
at which point a jet shoots out and heads towards the grounded collector. The higher the voltage
the faster the solution will be whipped away from the needle, accordingly to maintain a stable
electrospinning system the voltage must be matched to the federate. If the voltage used is too
high, multiple jets can be produced at the nozzle and if the voltage is too low, the excess solution
will build up and drip on the collector. For each solution a minimum voltage is needed to distort
the droplet into a Taylor cone, usually around 6kV. In flight, increasing the voltage has the effect
first of extending the fibres to make them longer and thinner and then of decreasing the flight time
to produce thicker fibres. The higher the voltage the thinner the fibres will be as they will have
more stretching force. An AC voltage can be used to give more aligned fibres.
40
3.1.2.2 Distance
The greater the distance between the tip and collector, the longer the jet will spend in flight and
the smaller the electric field strength will be. Potentially, thinner and thicker fibres can be
produced on increasing the tip to collector distance depending on the solution properties. While
the reduced electric field means far less stretching force i.e. thicker fibres, the jet elongation can
be sustained for a greater time period i.e. thinner fibres. As the solvent must evaporate before the
jet reaches the collector, a minimum distance is needed for successful electrospinning of each
solution.
3.1.2.3 Flow rate
As the flow rate is increased, fatter fibres and then beaded fibres, can be produced. High flow
rates are useful to raise productivity but must be supported by high voltage, solvent vapour
pressure and sometimes large tip to collector distances to ensure the fibres are laid down dry.
3.1.2.4 Nozzle area
The small orifice allows a smaller droplet with a corresponding higher surface tension which
necessitates using a higher voltage. This results in smoother thinner fibres. It has been observed
that a smaller nozzle area results in less clogging.
3.1.2.5 Humidity
Electrospinning in a humid environment causes pores to form on the fibre surface. As the solvent
vaporises, it cools the jet and water droplets can condense onto the surface. The pores grow in
depth and size with humidity until a saturation point is reached. Also, rate of evaporation of the
solvent is influenced by the humidity, the higher the humidity the slower the evaporation.
3.1.2.6 Temperature
Changes in temperature can affect the viscosity of the solution. If the polymer is dissolved in a
poor solvent, the chains are usually tightly curled up; heating can boost the solvating power and
unravel the chains allowing more overlap and entanglement that increases the viscosity. If
however the solvent is good and the polymer is well dissolved, on warming, the resistance
between chains will decrease, reducing the viscosity of the solution. The solvent vapour pressure
increases with temperature, reducing the fibre drying time. At higher temperatures, spinning can
be achieved with a higher voltage or shorter tip-to-collector distance.
3.1.2.7 Pressure
If the pressure is lowered below atmospheric level, the solution will be pulled out of the needle
faster and faster and bubbling will begin to occur until it is impossible to electrospin.
41
3.1.2.8 Atmosphere
The effect on the jet will usually be the dependent on the behaviour of the atmosphere under
electric field.
Freeze drying
Electrospinning is an effective way to combine AB and PEO in a well-mixed microstructure, but
it is slow, 1ml/hr flow rate of a solution with 6wt% solid content will produce 0.06g/hr sample,
and highly involved. Multi-nozzle electrospinning goes some way to increase the sample
production rate but is more technologically challenging and expensive. In order to produce large
amounts of sample quickly and cheaply while keeping the PEO and AB in the well mixed state
they enjoy in solution, freeze-drying was investigated. Freeze-drying [148] is a method where
the abundance of solvent can be evaporated while maintaining the solution structure and keeping
the AB stable.
In a freeze dryer [148] the material or solution mix is frozen and the solvent sublimed off with a
low vacuum. The freezing rate affects the water crystal size and the resultant particle size also,
the slower the freezing rate the large the water crystallites can grow. The effective freezing time
is influenced by the sample volume and the dish shape; the greater the radius from the sample
core to surface, the slower the freezing will be. While heat can be extracted from all directions,
water will only evaporate from the top of the sample and the further the water has to travel through
the sample the slower the evaporation.
The solutions were produced as normal, dissolving the PEO before the AB, and then poured into
a wide dish; a bigger surface area aids evaporation, and placed in a standard freezer at -5°C. Once
frozen, for a minimum of 12 hours at -5°C, the sample was transferred to the freeze dryer. Each
sample remained under vacuum for 48 hours.
Freeze drying is an ideal method to combine species soluble in water. Clay and AB composites
were also formed in this way. 5wt% clay was added to an AB-water solution and frozen before
the water was sublimed off. The clay content was kept constant and the amount of AB changed
to achieve the various fractions desired.
Dry mixing
AB was ground in a pestle and mortar for 15 minutes. PEO as-produced powder was used without
modification. The powdered AB and PEO were combined at the required ratio and mixed together
with a spatula.
42
A vapour transport method for intercalation into graphite
A two zone vapour transport method for intercalating metals into graphite is well known [26].
The graphite is outgassed under vacuum to remove impurities before being connected to a second
chamber in which the intercalant metal is liquefied. The metal vaporises, filling both chambers
and because of the overpressure, intercalates into the graphite. The stage number n produced is
related to the temperature difference between the graphite and metal containing chambers, the
smaller the difference the smaller the stage obtained.
A development on this where the staging is not controlled is to submerse the graphite in the molten
metal of the desired intercalant; a single zone transport method [33], [149]. The graphite,
Madagascan flake, was outgassed to remove impurities before being transferred, while inside an
argon glove box, to a stainless steel tube. The calcium was added, the tube sealed and the entire
ensemble put into a furnace. The tube was held under vacuum (~1.0×10-7 mbar) at 465°C for 10
days during this time the calcium vaporises and intercalates into the galleries.
Analysing samples
Scanning electron microscopy
Scanning electron microscopy (SEM) was used analyse the microstructure of the samples [150],
a Hitachi Tabletop Microscope TM-1000 enabled a quick first pass view of the sample and later,
high resolution images were obtained with a FE-SEM Hitachi S4000 and Jeol JSM-6480LV. The
samples were mounted on carbon tabs and platinum coated to prevent charge build up (note: no
coating was required for the TM 1000). To observe the fibre cross-section, the fibres were sealed
in adhesive tape that was then cut along the normal to the fibre length, while submerged in a liquid
nitrogen bath to keep the fibres brittle.
Compared to traditional light microscope an SEM can create clearer images of greater
magnification due to the smaller wavelength of the electrons.
Electrons are generated by thermionic emission from a filament, usually tungsten and the
accelerated with a positive electrical field to 1-30keV. The electron beam is focussed with
magnetic condenser lenses to a spot size of 1-10nm at the sample. A picture of the entire surface
is obtained by the raster motion of the beam, across the sample, controlled by scanning coils
responsible for the forward and back and the side to side movement.
43
Figure 17: Schematic of a scanning electron microscope.
The electrons interact with a surface over an interaction volume; some rebound because of their
high angle of incidence on the sample, others exchange energy leading to the release of secondary
electrons and x-rays. Secondary and backscattered electrons are used to create an image of the
surface, scan verses detection time make the picture three dimensional, while the x-rays can be
studied to determine the elemental composition. A photomultiplier is used to enhance the electron
signal.
Condenser
Lenses Double Deflection
Scanning Coils
Secondary
Electron Detector
Rotations and
Displacement
Electron Gun
Back Scattered
Electron Detector
Stage
Electron Beam
Anode
44
Fourier transform infrared spectroscopy
Fourier transform infrared (FTIR) spectroscopy was performed on some of the samples to analyse
the vibration states of the bonds [151]–[153]. A specific bond vibrational state, stretching,
bending, twisting etc. can be excited to its excited state by absorbing IR radiation of the correct
frequency. The frequency must match the energy difference between the ground and excited
states. A sample is subjected to a spectrum of IR radiation and the position of the absorptions
bands will indicate the energy of the motions inside the material. This allows bonds to be
identified and helps elucidate the structure of the material.
An IR spectrum is usually collected over a wavenumber from 4000 to 400 cm-1. Bond stretches
(functional groups) are observed at the high wavenumber end of the spectrum. The low
wavenumber section of the spectrum, from 1450cm-1 to 400 cm-1, is known as the fingerprint
region (every molecule has a distinct finger-print region pattern) and bending and twisting
vibrations are found here.
A bond is infrared active if the vibration results in a net change in dipole moment and, as dipoles
are vector additive, not all bonds give an IR signal. Carbon dioxide for example has IR active and
inactive stretches; 𝐶 ⃗⃗ ⃗⃗ = 𝑂 = 𝐶 ⃗⃗⃗⃗ gives a signal but 𝐶 ⃖⃗ ⃗⃗⃗ = 𝑂 = 𝐶 ⃗⃗⃗⃗ and 𝐶 ⃗⃗⃗⃗ = 𝑂 = 𝐶 ⃖⃗ ⃗⃗⃗, where the
dipole sum is zero, do not.
The position on the IR spectrum, �̅�, of a particular vibration can be defined by Hooke’s law,
�̅� =1
2𝜋𝑐√𝑘
𝜇𝑟
( 26)
where, k is the force constant of the vibration and 𝜇𝑟 is the reduced mass of the of the atoms on
either end of the bond. The reduced mas can be defined,
𝜇𝑟 =𝑚1𝑚2𝑚1 +𝑚2
( 27)
where, 𝑚1 and 𝑚2, are the masses of the atoms at either side of the bond. From this it is clear that
the adsorption band moves to higher wavelength as the masses of the atoms increase and/or as the
bond length decreases. The intensity of a particular vibrational band is principally due to the
magnitude of the change in dipole that occurs.
Attenuated total reflection Fourier transform infrared (ATR-FTIR) data were collected in
absorption mode from the post-heated materials at room temperature on a Bruker Optics Vertex
70 spectrometer with a 633 nm laser with 1cm-1 resolution. The peak positions were determined
by fitting Gaussian curves to the spectrum.
45
Foam tests
The materials were heated in an oil bath at 120°C to determine the foaming characteristic. All
foam tests were performed on pellets made in a small pellet press (see Figure 37). The initial
mass, height and diameter of the pellets were measured with callipers and the volume and density
calculated. After heating the pellets’ mass and size were measured again. The volume and mass
change due to the foaming could then be calculated for a particular pellet. If the pellet could not
be removed intact, the diameter was taken as the internal diameter of the test tube. Residue on the
inside of the tube wall was included in the final mass calculation and weight gain from the oil on
the test tube was also taken into account. No correction was made for the rounded profile of the
bottom of the test tube. The final height of the expanded pellet as measured with the callipers was
subject to several mm of error so it was not worth correcting for the rounded bottom of the test
tube.
Figure 18: AB containing pellets prepared for foam testing.
Thermogravimetric analysis and differential scanning calorimetry
Differential scanning calorimetry (DSC) is a thermoanalysis technique to study the phase changes
in materials with respect to temperature [154]. The sample temperature is raised at a constant rate
in a contained environment alongside a reference material and the difference in heat input,
required to keep the sample and reference at the same temperature is recorded. This is an
extremely useful technique for studying AB as melting events (endotherms) hydrogen release,
(exotherms) and phase transitions can all be readily identified.
Thermogravimetric analysis (TGA) measures the mass changes that occur in a material during
heating. It is an extremely useful accompaniment to DSC for understating changes that occurring
during a reaction and is often part of the same instrument.
A combined DSC/TGA produced by Mettler Toledo (TGA/DSC 1) was used for analysing AB
and the AB composites produced in this study (see Figure 19).
46
Figure 19: Schematic of a combined DSC/TGA instrument.
The effect of temperature on the samples was studied with both heat ramps (2°C min-1 between
room temperature and 200°C) and isothermal heating steps (80°C, 100°C, 120°C and 150°C).
Each sample was run three times to ensure the features observed are characteristic of the material
in general and the trace that best represented each sample is displayed in the graphs.
3.3.4.1 Validity
The differential temperature measurement, the output form the DSC, is sensitive to the heating
rate, sample size and shape and the gas flow rate [155]. Under ramp heating, a faster ramp rate
will lower the peak resolution and push up the peak temperature. AB undergoes several heat
induced changes (both exothermic and endothermic) below 200°C, therefore the ramp heating
rate was set low (2°C min-1) to ensure the peak features would be distinct. A larger sample will
take longer to react and this will lower the peak resolution. This means that under a ramp heating,
a larger sample will push the peak to higher temperatures while on an isothermal run, it will push
the peak to later times. As far as possible the sample masses were kept constant, however due to
the excessive foaming of AB and the small size of sample pans available, the mass of pristine AB
was limited to 1mg. As the composites had reduced foaming and AB content, higher masses
could be tested. For this reason caution should be taken when comparing the pristine AB and the
composites. Unreliability also comes from are the sample packing in the crucible, if the sample is
whole it will heat at a more consistent rate then if it is powdered. All the samples were squashed
to the base of the pan to ensure good contact. However, variation in the initial compression and
expansion during the test could change the sample connection to the crucible and still have an
appreciable effect. The argon flow through the chamber affects the heat control and the buoyancy.
As the DSC chamber heats so too does the gas inside i.e. argon. This means the argon density
decreases and reduces the buoyancy making the sample appear to gain mass. Convection currents
and the gas flow velocity can also impact the recorded mass and the heating of the sample.
47
The DSC and TGA signal from the pristine AB and the AB-PS material were both affected by a
ramping baseline shift. The TGA data slopes upwards and displays a slight increase before mass
loss begins significantly. The DAC data also slopes upwards and to enable clear interpretation of
the graph features a background quadratic curve was subtracted.
Mass spectrometry
To understand the dehydrogenation process of the AB composites, a mass spectrometer (mass
spec) was used to analyse the residual gases released during the decomposition of pristine AB and
the AB composites (see Figure 20). A quadruple residual gas analyser (RGA) (MKS Cirrus 2
Atmospheric pressure RGA) was connected to the out flow of the combined TGA/DSC.
A quadrupole mass analyser (QMS) works like a standard mass spec but instead of separating the
ions with a bending magnetic field the ions are filtered through a changing magnetic environment
[156]. Four conducting rods are arranged in parallel and a varying current applied to allow
specified mass/charge (m/z) particles to pass through to the detectors.
Figure 20: Schematic of a residual gas analyser (mass spec).
A vacuum is established in the mass spec. As the sample is heated in the DSC a steady stream of
argon is introduced to the sample chamber to transport the gaseous products into the mass spec.
Once inside a filament ionises the gas cloud by stripping away electrons, creating a plasma of
charged particles. Often molecules will be destroyed in the process and it is common also to
observe the fragments. The cloud is funnelled via the QMS into the ion detectors.
Table 8. The mass number assignment of the ionised species observed in the mass spec.
Mass Number
m/z 2 17 18 27 32 36 81
Assignment H2+ OH+, NH3
+ H2O+ B2H6+, BNHx
+ NH3BH3+ Ar+ B3N3H6
+
Filament
Gas input
Ionization chamber
Ion focus
Quadrupole mass filter
Ion detector
48
The mass spec analysers can detect species of mass to charge ratio from 1 to over 100. I have
chosen to consider the classic gaseous products and impurities released on heating AB, hydrogen
(m/z = 2), ammonia (m/z = 17), diborane (m/z = 27) and borazine (m/z = 81). And to ensure no
unreached AB is thrown off the material the AB (m/z = 32) has been observed. As water is also a
known fuel cell contaminant and AB is known to be hydroscopic, the water signal water (m/z =
18) has also been monitored. In addition the argon (m/z = 36) carrier gas flow level has been
followed as it will affect the partial pressure of the ionised reaction products and therefore the
intensity of the data readings. Finally, the list here is far from conclusive as only a few mass
numbers from the whole possible range have been observed; furthermore each detected mass
number can represent a variety of different ions, for example m/z = 17 can represent ammonia or
a hydroxyl [53]
The data presented has been processed from its initial raw form to clarify the important features
and give fair consideration to the errors. Each composite has been put through the same heating
test three times on the same day to ensure consistency. The data sets have been normalised to the
mass of AB in the sample and the argon gas flow so what begins as a reading of current is
transformed into arbitrary units (a.u.) per gram (g-1). The argon gas flow rate is taken from the
average value of the m/z = 36 trace from 5 to 9 minutes, the interval before heating commences.
Because the data are noisy mathematical smoothing via the Savitsky-Golay method has been
employed [157]. Savitsky-Golay smoothing uses a least squares fitting and sustains peak features
better than adjacent averaging methods. It calculates a polynomial regression, 𝑝(𝑛𝑓), around the
central point, 𝑛𝑓, of a moving envelope of points, 𝑁 = 𝑛0, 𝑛1…𝑛𝑛 and outputs a new smoothed
value for that point, 𝑛𝑔.
𝑝(𝑛𝑔) = ∑𝑛𝑔𝑛𝑓𝑘
𝑁
𝑘=2
( 28)
A second order polynomial, 𝑘 = 2, gave a suitable fit in all cases. However, the value of N (=
2M+1) was varied between 50 and 500 depending on the trace, typically the data for hydrogen
(m/z = 2) used n=50 and borazine (m/z = 81) used n=5000. The fit is decided by minimising the
least squares error, E, of the new points to the original values,
𝐸 = ∑ (𝑝(𝑛𝑔) − 𝑛𝑛 )2
𝑀
𝑛=−𝑀
( 29)
In all cases, the goodness of smoothing was assessed by eye to ensure no obvious features had
been erased, an example of a particularly noisy ammonia (m/z = 17) signal is below in Figure 21.
49
Figure 21: Savitsky-Golay smoothing, of ammonia (m/z = 17) signal where N = 500.
From the three normalised curves of each mass number detected, a mean curve and error (standard
deviation) was calculated. These graphs are used to show the changes in the reaction over time.
Further, the peaks are integrated and normalised to the hydrogen signal level at maximum times.
This is used to show the relative levels of impurities to hydrogen in the various materials.
3.3.5.1 Validity
Finally, it is important to discuss the validity of the mass spec data. To ensure precision three data
traces form each sample have been combined to give a mean trace (the values) and standard
deviations (the errors). In Figure 22 the data form two separate AB samples is shown, AB1 and
AB2. Both AB samples have been sourced form Sigma Aldrich, are 97% pure and were tested in
their as-received state. The two samples were tested on the same equipment several months apart.
The small error bars in the hydrogen trace (light background) show that data from the three traces
of each sample are well correlated; AB2 is more similar to itself than AB1.
Looking below to the impurity graphs in Figure 22, AB2 seems to release far more ammonia,
diborane, borazine and water than AB1. Both sets of data have been processed in the same ways,
normalising to the AB mass, the argon flow and have undergone the same smoothing technique.
By normalising to the nominal mass of AB in the samples it has been assumed that the output
signal from the detectors varies linearly with the gas output from the samples. This has not
actually been verified, indeed the mass spec manufacture suggested that the ratio between gas
input and signal output could vary. The data are normalised to the carrier gas stream (argon)
because logically a higher, faster throughput will result in a higher signal. However, there is no
actual information to suggest that this is the case. Perhaps a lower, slower gas flow would actually
result in a higher signal as the gas would not be rushed past the detectors as quickly. Secondly, it
has been assumed that all mass fragments here are detected equally. It is possible that there is
60 80 100 120 140 160 180 200
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
Am
monia
(m
/z=
17)
signal
, ar
bit
ry u
nit
s
Temperature, oC
Normalised data
SG smoothing
50
some variable sensitivity to the different mass fragments; some may be more equal than others
[158]. Also some mass fragments may ionise more easily than others.
The differences between the impurity levels in Figure 22 are probably due the samples being run
several months apart. Tests performed on the same day show greater similarity then those taken
a year apart, due to changes in the environment, equipment and materials.
Normalising to the mass of AB is valid as the detection level can be approximated as linear when
the samples are of a similar mass. Sequences of samples are well correlated so this is a credible
theory. (Consider the ammonia traces form the 25wt%, 50wt% and 75wt% AB in clay (imogolite)
in Figure X in chapter 6.) However, when the mass difference between the samples is large
(comparing AB in clay to pristine AB in chapter 6) the intensities of the integral plots should not
be trusted. In many cases the pristine AB has been plotted alongside the composites as comparing
the graph profiles can be useful.
51
Figure 22: Cumulative integrals of the heating ramps of two pristine AB samples (AB1 and AB2) tested
several months apart. The data were normalised to the hydrogen (m/z = 2) level at 190ºC and errors in the
data are shown as the lighter background. The increase in impurities observed in AB2 is likely due to the
time separation between the two tests. This graph has been plotted to show care must be taken when
comparing impurity levels from different samples.
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
40 60 80 100 120 140 160 180
m/z
=2
7,
no
rmal
ised
to
hy
dro
gen
sig
nal
Ammonia
m/z
=1
7,
no
rmal
ised
to
hy
dro
gen
sig
nal
40 60 80 100 120 140 160 1800.00
0.01
0.02
0.03
0.04 BorazineDiborane
Pristine AB
Pristine AB 2
40 60 80 100 120 140 160 1800.000
0.001
0.002
0.003
0.004
0.005
0.006
0.007
40 60 80 100 120 140 160 180
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
m/z
=8
1,
no
rmal
ised
to
hy
dro
gen
sig
nal
Water
m/z
=1
8,
no
rmal
ised
to
hy
dro
gen
sig
nal
Temperature, 0C
Temperature, 0C Temperature,
0C
Temperature, 0C
40 60 80 100 120 140 160 180
0.0
0.2
0.4
0.6
0.8
1.0
1.2
m/z
=2
, n
orm
alis
ed
Temperature, 0C
Pristine AB
Pristine AB 2
1
52
3.3.5.2 Activation Energy
The activation energy of the dehydrogenation of the AB can be determined using the Avramie-
Erofeyev model [159]. The isothermal hydrogen release is sigmoidal as shown in Figure 22 and
specifies a nucleation, growth and saturation process [160]. The growth phase of the reaction,
moves outward in all directions from the nucleation point [54]. The Avrami-Erofeyev equation
is reserved for solid phase reactions. While AB melts during the dehydrogenation the actual
molecules maintain their approximate position and so the reaction can be modelled as a solid.
Figure 23: The mass loss (top) of the 75AB-25PEO electrospun fibres when subjected to heat steps of
150°C, 120°C and 100°C in the combined TGA/DSC (bottom). In the 150°C and 120°C steps the
majority of the hydrogen has been lost before the maximum temperature was reached. Additionally the
samples never reach the set temperature. Therefore this loss data are inappropriate for analysing the
activation energy of AB.
Close examination of the TGA and mass spec data revealed several issues that meant it could not
be used for this type of analysis. The step heats are not really step heats as the actual sample
temperature lags so far behind the set temperature. The instrument reaches the target temperature
in ~2 minutes (a rate of 75°C/min) but the sample temperature lags behind by approximately 20
minutes (see Figure 23). Also, the final temperature attained is several degrees below the target.
In the 150°C and 120°C runs the majority of the hydrogen release occurred before the maximum
temperature was reached. This means that most of the hydrogen released during the step heats is
10 20 30 40 50
40
60
80
100
120
140
0.4
0.6
0.8
1.0
10 20 30 40 50Time, min
Tem
per
ature
, oC
Time, min
Set temperature
Sample temperature
Mas
s lo
ss,
%
75AB-25PEO 1500C step
75AB-25PEO 1200C step
75AB-25PEO 1000C step
53
actually released on a fast 20 minute ramp heating. For this reason it was not possible to calculate
the activation energy of the hydrogen release from AB in the composites.
X-ray diffraction
A diffraction pattern can be describes as the interference pattern that occurs if a wave encounters
the regular lattice of a crystal [161], [162]. The recurring dimensions of the lattice, d, can be
obtained by relating the angle of the incident beam, 𝜃, via the Bragg approximation,
𝑛𝜆 = 2𝑑𝑠𝑖𝑛𝜃 ( 30)
where n is an integer representing the scattering order and 𝜆 is the wavelength. Sharp peaks occur
with constructive interference and these diminish as the lattice becomes disordered. A diffraction
pattern of a material gives access to the repeat distances inside and is therefore ideal for
determining the crystal structure. X-rays are ideal for exploring materials as the wavelength is
comparable to interatomic spacings.
Traditional methods of obtaining a crystal diffraction pattern with x-rays require a single crystal
that must be correctly aligned to the beam to expose the plane of interest. Diffraction from other
repeat distances in the crystal can be only be accessed by turning the crystal to reveal a new plane
to the beam. Making and aligning a single crystal is challenging at best and a second method
designed for powders is more appropriate. The Debye-Scherrer method uses the distribution of
the random crystallite orientations in a powder sample to present multiple planes to the beam at
the simultaneously. The sample is fixed to the sample holder and revolves to achieve an average
distribution of the scattered intensity.
Electrons are produced by thermionic emission off a cathode and accelerated towards a cooled
metal target, often copper as it has a high thermal conductivity. On collision the electrons are
decelerated in the material which results in the emission of Bremsstrahlung photons – a
continuous spectrum of x-rays. During this process the inner shell electrons in the atoms are knock
out and when an outer shell electron drops down to replace the displaced one, an x-ray is emitted.
If the electron drops into a K shell this is called K-radiation, Cu Kα X-rays are particularly intense
peaks in the spectrum and are ideal to be used in a diffraction experiment. Once produced the X-
rays of the required wavelength are selected by diffracting off a graphite lattice. The x-rays are
detected by a NaI scintillator with a combined photomultiplier to strengthen the signal.
Diffraction patterns were collected to study the structure and composition of the materials. A
Rigaku SmartLab diffractometer with Cu Kα radiation of wavelength 1.54Å between 10 and 110°
with a scan rate of 1.50 deg/min and resolution of 0.0002° was used. Lower resolution patterns
54
were also collected with a high and a Phillips X’pert Xray diffractometer with Cu Kα radiation of
wavelength 1.54Å between 5 and 75° with a scan rate of 1deg/min and resolution of 0.02°.
Nuclear magnetic resonance spectroscopy
Nuclear magnetic resonance (NMR) is an extremely useful technique to study the structure,
dynamics, reaction states and the chemical environments of magnetic nuclei.
Isotopes with an odd number of protons and/or neutrons have a magnetic moment (μ) and angular
momentum. Boron has five protons and has a magnetic moment that can be explored by NMR
while 16O and 12C are isotopes with an even number of protons and neutrons and therefore do not
have a magnetic moment. Protons and neutrons all have spin of ±1/2. The magnetic moment is
related to the nuclear spin quantum number (S),
μ = S ∙ γ𝑔
where γ𝑔 is the gyromagnetic ratio; the ratio of the magnetic dipole to the angular momentum.
There are 2S+1 angular momentum states from –S to +S in integer steps.
The nuclear magnetic moment of a nucleus will align with an external magnetic field either
parallel or antiparallel. The nuclei magnetic moment will precess around the direction of the field.
Aligning parallel to the field is preferred then aligning against it. The larger the magnetic field,
the larger the energy difference between the two states.
In nuclear magnetic resonance (NMR) spectroscopy the nuclei are subjected to a magnetic field,
up to 20tesla, to align the spins parallel and antiparallel to the field. Then radio frequency waves,
energise the lower energy (parallel aligned) nuclei which flip into the higher energy (antiparallel
aligned) state. When the frequency is removed the nuclei relax back to their initial states
perturbing the magnetic field. This perturbation is called the resonance and is gives the measured
signal.
The frequency, ν, at which a resonance can occur is related to the external magnetic field, 𝐵0,
𝜈 =
𝜇𝐵0ℎ𝑆
( 31)
where 𝑆 is the spin quantum number and ℎ is Planks constant. Thus if a high field is used to
polarise the spins a higher frequency electromagnetic radiation is required.
Electrons have a magnetic moment as well as protons and neutrons. The electron spin opposes
the nuclear spin and can shield the nucleus from an external magnetic field. High shielding will
55
shift the resonance to lower frequency while low shielding will shift the resonance to higher
frequency.
NMR spectroscopy is ideal for studying AB as AB contains three NMR active nuclei, boron,
nitrogen and hydrogen. While both 11B and 10B are NMR active and have spins of S = 3/2 and 3
respectively, however, 11B NMR was performed on AB as 11B has an 80% abundance and gives
a clearer signal. Time resolved studies of AB and the AB composites during the dehydrogenation
could be performed at 85°C. Nitrogen has two naturally occurring isotopes 14N and 15N with
abundance 99.6% and 0.4% respectively. While both isotopes are NMR active, 14N is hampered
by quadrupole interactions which broaden the peak and can render it unobservable under high
resolution. Therefore, 15N NMR was performed on the AB samples. As the natural abundance of
15N is very low, the spectra must be left to accumulate for several hours.
In situ solid-state 11B NMR was performed on the pristine AB and the 50AB-50PEO electrospun
material while being heated at 85°C. NMR magic-angle spinning (MAS) spectra were recorded
at 128.3 MHz using a Varian VNMRS 400 spectrometer and a 4mm (o.d.) rotor. Spectra were
obtained using cross-polarisation with a 1.0s recycle delay, 40ms contact time and at a sample
spin-rate of 10kHz. Between 20 and 100 repetitions were accumulated. Spectral referencing was
performed with respect to an external sample of F3B-OEt2. Additionally, 15N NMR was performed
on the initial and final products (after heating in oven at 85 °C for 6 hours) at ambient temperature.
The spectra were recorded at 40.527 MHz using cross-polarisation with a 5.0s recycle delay, 40ms
contact time and at a sample spin-rate of 6.8 kHz. The data were referenced to neat nitromethane.
Introduction to neutron scattering
Neutron scattering is a valuable tool for exploring condensed matter systems. Both structure and
motions can be studied as the neutron wavelength, λ, is akin to interatomic spacings. Being
uncharged, neutrons can penetrate far into matter where they interfere directly with the nucleus.
Neutrons have mass and accordingly, can scatter inelastically; the direct nuclear interaction
ensures intensity is maintained at high scattering angle. Interactions are irregular with respect to
nuclear size and isotope allowing isotope substitution to gather more information.
The energy of a neutron 𝐸𝑛, is described
𝐸𝑛 =𝑝2
2𝑚𝑛=𝑚𝑛𝑣
2
2=𝑚𝑛2(𝐿
𝑡)2
( 32)
where 𝑝 is the momentum, 𝑚𝑛 is the neutron mass equal to 1.675·10-27 kg, v is the velocity, t is
the time and L is the distance travelled. Using the de Broglie relationship
56
𝑝 = ℏ�⃗� ( 33)
where �⃗� is the wavevector of the neutron propagation and ћ is the reduced Planks constant equal
to 6.582×10−16 eV·s. We also note that
�⃗� =2𝜋
𝜆
( 34)
where λ is the neutron wavelength.
When a neutron, with a propagation vector, collides with a nucleus it will undergo a momentum
change, ∆𝑝 = �⃗� = �⃗� 𝑖 − �⃗� 𝑓, where �⃗� 𝑖 and �⃗� 𝑓 are the initial and final wavevector of the neutron
respectively. The neutron can also gain or lose energy to the sample, ∆𝐸 = ℏ𝜔 = 𝐸𝑖 − 𝐸𝑓 =
ℏ2
2𝑚(𝑘𝑖 − 𝑘𝑓), where 𝐸𝑖 and 𝐸𝑓 are the initial and final energies of the neutron (see Figure 24).
The scattering process is described as elastic when the neutron energy remains constant and
inelastic when there is a change. Analysing these changes allows us to understand the structure
and motions inside materials.
𝑘𝑖 = 𝑘𝑓 𝑁𝑒𝑢𝑡𝑟𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 } 𝐸𝑙𝑎𝑠𝑡𝑖𝑐 𝑠𝑐𝑎𝑡𝑡𝑒𝑟𝑖𝑛𝑔 → 𝑄 =4𝜋𝑠𝑖𝑛𝜃
𝜆
( 35)
𝑘𝑖 > 𝑘𝑓 𝑁𝑒𝑢𝑡𝑟𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦 𝑙𝑜𝑠𝑠
𝑘𝑖 < 𝑘𝑓 𝑁𝑒𝑢𝑡𝑟𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦 𝑔𝑎𝑖𝑛}
𝐼𝑛𝑒𝑠𝑡𝑎𝑠𝑡𝑖𝑐 𝑠𝑐𝑎𝑡𝑡𝑒𝑟𝑖𝑛𝑔 →
𝑄 = √2𝑚
ℏ2[2𝐸𝑖 − ℏ𝜔 − 2√𝐸𝑖(2𝐸𝑖 − ℏ𝜔) ∙ 𝑐𝑜𝑠2𝜃]
( 36)
Figure 24: The momentum and energy changes associated with elastic and inelastic scattering. 𝑘𝑖 and 𝑘𝑓
are the initial and final wavevector of the neutron respectively.
A neutron beam at a facility will combine a distribution of neutron energies these can are generally
described according to their temperature (see Table 9).
ki
ki = k
f
Q
ki > k
f
ki < k
f
kf
kf
ki
Detector
Sample
Monochromator
2θ
2θ
57
Table 9: Neutrons according to temperature [163]
Neutron Energy (meV) T (K) λ (Å)
Hot 100-500 1000-6000 0.4-1
Thermal 5-100 60-1000 1-4
Cold 0.1-10 1-120 4- 30
The neutrons that have been used in the experiment were thermal neutrons. The data were
collected on the Iris beam line at Isis, Rutherford Appleton Laboratory, UK.
In a scattering experiment we measure the partial differential cross-section. This is the fraction of
neutrons of incident energy E scattered into a solid angle element, dΩ with an energy transfer
element, dω. Here, 𝜎 = 4𝜋𝑏2 is the total scattering cross section of the atoms in the sample and
b represents the scattering interaction of the particular nucleus (see Figure 25). 𝑆(�⃗� , 𝜔) is the
dynamic structure factor, the total scattering with respect to Q and ω.
Figure 25: Geometry of a scattering experiment [164]
𝑑2Ω
𝑑Ω𝑑𝜔=𝜎
4𝜋.|�⃗� 𝑓|
|𝑘𝑖⃗⃗ ⃗|𝑆(�⃗� , 𝜔)
( 37)
As well as elastic and inelastic, a scattering interaction is also described as coherent or incoherent.
Coherency, 𝑆𝑐𝑜ℎ(𝑄,𝜔), arises from interference scattering off a regular lattice within the
material. Elastic coherent scattering can be used to pinpoint the locations of atoms while inelastic
coherent scattering is ideal for studying excited states such as phonons. Incoherent scattering,
𝑆𝑖𝑛𝑐(𝑄,𝜔), is the result of neutrons scattering independently off atoms. Whereas elastic
incoherent scattering is usually isotropic and seen as a diffuse background, inelastic incoherent
scattering gives information about uncorrelated motion of nuclei such as diffusion and vibration.
58
Each nuclei has a coherent scattering cross section, 𝜎𝑐𝑜ℎ, and an incoherent scattering cross
section 𝜎𝑐𝑜ℎ such that,
𝜎𝑐𝑜ℎ = 4𝜋𝑏𝑐𝑜ℎ2 and 𝜎𝑖𝑛𝑐 = 4𝜋𝑏𝑖𝑛𝑐
2 ( 38)
Below in Table 10 are the scattering cross sections of the elements used in the experiments in this
thesis. Dynamics studies of hydrogen or hydrogen containing groups are facilitated by the high
incoherent cross section of the proton. The variance between hydrogen and deuterium enables
contrast studies to be performed between the protic and non protic areas as well as absolute
scattering experiments.
Table 10: Scattering lengths and cross sections of the nuclei observed in experiments in this thesis.
Species bcoh/fm binc/fm σcoh/barn σinc/barn
H −3.74 25.27 1.76 80.26
D 6.671 4.04 5.592 2.05
C 6.64 0 5.55 0.00
Ca 4.90 0 2.78 0.05
N 9.36 2.0 11.01 0.5
B 6.65 -1.3 5.56 0.21
O 5.803 0 4.232 0.00
S 2.804 0 0.988 0.00
The total scattering, is the sum of the coherent and incoherent parts, 𝑆(�⃗� , 𝜔) = 𝑆𝑐𝑜ℎ(𝑄,𝜔) +
𝑆𝑖𝑛𝑐(𝑄, 𝜔) and depending on the identity and number of atoms being irradiated, a scattering
intensity will be measured. However, to determine anything useful about the system the atoms
must be correlated, i.e. must relate meaningfully to each other. Correlation between atoms can be
described as regularity within the sample structure, such as the repeat distance between atoms in
crystal lattice sites, distribution of molecules in a liquid or the unique rotational energy of a
chemical group. The time-dependent self-pair correlation function 𝐺(𝒓, 𝑡) is a theoretical
description of the probability of encountering atoms at position r and time t with reference to the
origin, r = 0 and t = 0. It allows 𝑆(�⃗� , 𝜔) to be Fourier transformed into a real space and time
description of the positions, motions and energetics of the atoms.
𝑆(�⃗� , 𝜔) =
1
2𝜋ℏ∬𝐺(𝒓, 𝑡)𝑒𝑥𝑝{𝑖(𝑄 ∙ 𝑟 − 𝜔𝑡)} 𝑑𝒓𝑑𝑡
( 39)
The reader can extend this basic introduction of neutron scattering with this excellent book by Willis et al
[165].
59
IRIS
Crystal diffraction patterns and spectroscopic studies of Ca-NH3 GIC were performed on the IRIS
instrument at the ISIS Neutron Scattering Facility at the Rutherford Appleton Laboratory in the
UK. IRIS is an indirect geometry and is a time of flight spectrometer (see Figure 26). IRIS has
cold neutrons, when these arrive they are moderated, passed along a guide and through two
choppers which are used to define the wavelength distribution at the sample. As a spread of
neutron energies is present the initial flight time, t1, along the initial flight path, L1, will be
different. However, an analyser crystal ensures that only neutrons of a defined energy that satisfy
Bragg’s Law reach the detectors.
Figure 26: Schematic of IRIS indirect time of flight neutron spectrometer
The final energy, 𝐸𝑓, can be calculated by combining equations, x and y
𝐸𝑓 =𝑚𝑛2(𝐿2𝑡2)2
= 𝑚𝑛𝑣
2
2=𝑝2
2𝑚𝑛=
1
2𝑚𝑛(ℎ
𝜆)2
=1
2𝑚𝑛(
ℎ
𝜆2𝑑𝑎𝑠𝑖𝑛𝜃)2
( 40)
Between the sample and the detector the flight length of the neutron, L2 is accurately known
therefore the time, t2, to make the journey can be calculated.
𝑡2 =𝑚𝑛𝐿2𝑑𝑠𝑖𝑛𝜃
ℎ
( 41)
Should inelastic scattering occur, a spread of neutron energies and therefore arrival times at the
detector will be observed. The energy change, ∆𝐸, of the neutron can be ascertained,
∆𝐸 = 𝐸𝑖 − 𝐸𝑓 =
1
2𝑚𝑛 [(
𝐿1𝑡 − 𝑡2
)2
− (𝐿2𝑡2)2
]
( 42)
IRIS has detector coverage for a Q range of 0.4 to 3.7Å-1 for spectrometry and crystallography,
with additional detectors for purely diffraction experiments, this case the scattered neutrons
L1,t
1
Detector
Sample
Monochromator
ϕ
θ
L2,t
2
Choppers
Analyser Crystal
60
bypass the analyser crystal. Diffraction data can be collected on IRIS in in sync with a scattering
experiment or the instrument can be setup for a purely diffraction experiment. During the
experiment diffraction data were collected over a d-spacing range from 3.4 to 7.4Å using the IRIS
settings d-range 3, 4 and 5. Data presented with a smaller d-spacing has been collected during the
scattering experiments and then collated onto the graph. The scattering data were collected over
an energy range of -0.3 to 1.2 meV with pyrolytic graphite crystals. The neutron energy of 1.845
meV affords an energy resolution of 17.5 µeV. The instrument resolution was defined with a
vanadium annular cylinder as vanadium is an isotropic, purely elastic scatterer.
The total spectrum obtained from a neutron experiment is composed of elastic, inelastic and quasi-
elastic regions as shown in Figure 27. Quasi-elastic neutron scattering (QENS), inelastic
scattering close to the elastic line, can access timescales of motion of 1ps – 1ns and therefore the
movements both rotation and translational.
Figure 27: Schematic of the neutron scattering energy spectrum.
The incoherent dynamic structure factor, 𝑆𝑖𝑛𝑐(𝑄,𝜔), is taken from the QENS and is a function of
the translational, 𝑆𝑖𝑛𝑐𝑡𝑟𝑎𝑛𝑠 (𝑄, 𝜔) and rotational, 𝑆𝑖𝑛𝑐
𝑟𝑜𝑡 (𝑄,𝜔) incoherent structures and the mean
square displacement ⟨𝑢2⟩. The convolution of the three terms can be expressed:
𝑆𝑖𝑛𝑐(𝑄, 𝜔) = 𝑒
−1
3𝑄2⟨𝑢2⟩𝑆𝑖𝑛𝑐𝑡𝑟𝑎𝑛𝑠 (𝑄,𝜔)𝑆𝑖𝑛𝑐
𝑟𝑜𝑡 (𝑄,𝜔), ( 43)
where ℏ𝜔 is the energy transfer and ℏ𝑄 the momentum transfer of the neutron. At low Q values,
where, compared to the mean time between atomic collisions the times are long, the diffusion
Elastic scattering
Quasi-elastic scattering
Energy transfer, Q
0
Resolution width
Inelastic scattering
Neu
tron c
ounts
FWHM
Quasi elastic line
(sum of QENS and
resolution function)
ℎ𝜔2𝜋 < 0 ℎ𝜔
2𝜋 > 0
61
process is ruled by the solution of Fick's Law and results in a translational incoherent scattering
function of the form:
𝑆𝑖𝑛𝑐(𝑄, 𝜔) =1
𝜋
ℏ𝐷𝑄2
(ℏ𝐷𝑄2)+(ℏ𝜔)2 ,
( 44)
where D is the diffusion coefficient. Thus, the translational incoherent scattering function is a
single Lorentzian function, with the full width half maximum (FWHM),
Λ(𝑄) = ℏ𝐷𝑄2. ( 45)
Further information about the diffusion process can be gained from the QENS by considering the
higher Q values as deviation from the Fickian model at high Q would suggest jump diffusion.
Several models have been considered for the system particularly the Singwi-Sjölander [166] and
the Hall-Ross [167] and Chudley-Elliot [168] (see Table 11). While evenly distributed, the
calcium in the layers do not occupy a perfect lattice, consequently a distribution of the ammonia
jump lengths is probable. All the models afford good fitting and give similar values of l and 𝜏.
The Hall-Ross model has been selected in this case.
Table 11: Jump diffusion models, l is the jump length and 𝝉 is the average translational residence time of
the diffusing particle.
Model Chudley-Elliot Singwi-Sjölander Hall-Ross
Λ(𝑄) =
ℏ
𝜏(1 −
sin (𝑄𝑙)
𝑄𝑙)
ℏ𝐷𝑄2
1 − (DQ2𝜏)
ℏ
𝜏[1 − 𝑒𝑥𝑝(−Q2D𝜏)]
Jump
characteristic
Finite Exponential distribution Gaussian distribution
The diffusing particle is characterised by the diffusion coefficient is found from:
𝐷 =𝑙2
6𝜏 .
( 46)
where l is the jump length and 𝜏 is the average translational residence time of.
The rotational incoherent dynamic scattering function describes the rotation on the surface of a
sphere of radius R;
62
𝑆𝑖𝑛𝑐(𝑄,𝜔) = 𝑗02(𝑄 ∙ 𝑅)𝛿(𝜔) +∑(2𝑖 + 1)𝑗𝑖
2
∞
𝑖=1
(𝑄 ∙ 𝑅)
×1
𝜋
𝐷𝑟𝑜𝑡𝑖(𝑖 + 1)
(𝐷𝑟𝑜𝑡𝑖(𝑖 + 1))2 + (ℏ𝜔)2
( 47)
Where 𝐷𝑟𝑜𝑡 is the rotational diffusion constant and 𝑗𝑖(𝑄 ∙ 𝑅) is the ith spherical Bessel function.
In the context of the system under consideration only the first two terms (i=0,1) are required as
the higher order terms only have a significant contribution at larger momentum transfers then
those probed in this experiment. Within the scope of this series truncation the (Q independent)
FWHM is given by:
Λ(𝑄) = 2𝐷𝑟𝑜𝑡 =ℏ
3𝜏𝑟𝑜𝑡, ( 48)
where 𝜏𝑟𝑜𝑡 is the rotational dynamic correlation time.
Quasi-elastic neutron scattering (QENS), inelastic scattering close to the elastic line, can access
timescales of motion of 1ps – 1ns and therefore the movements both rotation and translational, of
the NH3 molecules can be tracked.
Modes
The data has been analyzed with MODES v3 a program designed specifically to process data from
the IRIS beam line [169]. MODES facilitates data binning and background subtraction, it displays
diffraction data and fits inelastic peaks with a combination of Gaussian and Lorentzian curves.
63
4 Ammonia borane and polyethylene oxide composites
This chapter describes the study on the AB and polyethylene oxide (PEO) composites. Both
making and testing the composites is explained. Here, PEO is shown to have a positive effect on
the dehydrogenation of AB.
Electrospinning polyethylene oxide
PEO has previously been electrospun with a range of solvents, including water, ethanol and
acetone, as part of a polymer blend or unaccompanied for a variety of applications including
filters, wound dressing and microelectronics [107], [132], [170]. Many solutions of various
concentrations of distinct PEO molecular weights have been processed under a range of spinning
parameters. However, considering the large number of changeable electrospinning parameters,
coupled with equipment and environment variables, it was deemed necessary to experiment with
various solutions as opposed to copying from literature sources.
At the time of writing PEO is cheaply and abundantly available in a wide range of molecular
weights. A selection of molecular weights between 400,000g/mol (4k) and 8,000,000g/mol (8M)
were purchased for testing. Deionised water was chosen as the solvent; it has been used
extensively for electrospinning PEO in the past, is benign and is a co-solvent for both PEO and
AB. PEO in water is termed PEO-H2O solution, and PEO in acetonitrile is termed PEO-ACN.
Table 12: An illustration of the electrospinning parameters tested, to determine the ideal conditions to
produce well defined PEO fibres from PEO in water solutions.
PEO Mw
g/mol
Conc.
wt%
Flow rate
ml/hr
p.d. keV Observations
400k 5 1 10 - 30 Spraying not spinning, no fibres
900k 5 1 10 - 30 15-17keV Spinning.
2M
2 1 - 5 9 - 16 Unstable
2.5 1 16 Damp fibres
3 1 14 - 16 Shiny fibre mat
4M 1 1 - 4 13 - 46 1ml/hr, 13-17keV spinning, but wet
8M
1 0.25 - 1 12 - 15 0.25 ml/hr, 12keV stable, but wet
2 2 10 - 30 14-16keV Spinning
The ideal system would be robust and have a high fibre production rate (flow rate multiplied by
concentration). The low molecular weight PEO solutions, 400 and 900k, were unsuitable as they
had a tendency to spray or to generate multiple jets from the droplet instead of spinning (see Table
12). The low viscosity and surface tension of the 400 and 900k PEO-H2O solutions seem unable
to support the charge necessary to eject a filament at a rate that prevents dripping. The high Mw
64
PEO solutions, using 4M and 8M, were also unsuitable, as they were extremely viscose at low
concentration, so a high solid content could not be achieved. The 2M PEO solutions were most
suitable as they could be stably spun with a range of concentrations and conditions; the 3wt%
solution is the most appropriate having the highest polymer content and was used as a base for
the AB PEO solutions.
Two issues arose. Firstly, the fibres adhered strongly to the aluminium foil collector and were
problematic to remove and secondly, the fibres were still damp when they reached the collection
plate. Replacing the aluminium sheet with grease proof paper, simplified the fibre collection but
the fibres were still too damp. In literature, dry fibres can be achieved by increasing the jet flight
time, lowering the humidity and raising the temperature of the spinning rig environment or
replacing the solvent with another of a higher vapour pressure [107]. The distance between the
nozzle and collector was already large (30cm) and maintaining a low humidity was problematic,
as the chamber leaked. Additionally, higher spinning temperatures were not possible as AB is
temperature sensitive. Therefore alternative solvents were considered.
3wt% PEO was added to a wide range of solvents (Figure 28) however, most were unsuitable.
PEO did not dissolve in ethanol. When acetone was added to PEO a white solid was produced.
PEO with DMA and DMSO produced a solid opaque gel. PEO in xylene, dichloroethane or
toluene resulted in a clear to opaque gel. DMF was a good solvent for PEO but the tube became
blocked during spinning. Acetonitrile, ACN, offered a viable alternative being a good solvent for
PEO and having a high vapour pressure.
Figure 28: 3% PEO (2M) in a variety of solvents.
To determine the effect of the ACN on the PEO–H2O solution, three solutions containing 0, 50
and 100% ACN as solvent were electrospun and compared (see Table 13). The ACN reduces the
Acetone, white solid Dimethylacetamide, white solid gel
pXylene, opaque solid gel Toluene, opaque solid gel
Dichloroethane, soft gel Water, cloudy
Ethanol, PEO powder grains remain
65
conductivity and the viscosity of the end solution as compared to water. The change in
conductivity is due to the difference in conductivity of the pure solvents, ACN being lower then
H2O. The drop in solution viscosity suggests the ACN is a better solvent. Either, the ACN is the
preferred solvent to H2O and the polymer chains are better lubricated and their reptation is
enhanced, or the ACN is a poor solvent and the polymer chains, although dissolved, remain curled
and the shorter end-to-end distance is responsible for the drop in viscidity.
Table 13: Electrospinning parameters of PEO (2M) solutions in H2O, in ACN and in a 50:50 mix
Solvent Conc.
wt%
Conductivity,
µSv/cm ±1
Viscosity high
shear, cP ±10
Flow,
ml/hr
p.d.
keV Observations
H2O 3 76 1760 1 10-20 12keV – flat
shiny mat
50ACN-
50H2O 3 18 1690 1 10-20
12keV – fluffy
pile
ACN 3 1 785 1 10-20 Spins fine
The carrier solvent used (H2O or ACN) dramatically affects the morphology of the electrospun
PEO fibres as shown in Figure 29. The SEM images in Figure 29 show that the PEO produces
thin smooth fibres while the PEO-ACN solution produced fibres which are wide and rough. The
PEO-H2O-ACN solution resulted in the most interesting morphology, reminiscent of beads on a
string.
Figure 29: PEO electrospun fibres with solvents H2O, ACN and an equal mix of the two.
The fibres were electrospun under similar conditions in the same eletrospinning rig, therefore the
differences observed are likely to originate from the solutions themselves. The PEO-H2O solution,
being highly conductive, is very responsive to the electric field, and would consequently
experience a high stretching force. The low vapour pressure of H2O will keep the jet fluid allowing
greater elongation to occur before drying (the fibres were damp on the collection plate) leading
PEO-H2O PEO-H2O-ACN PEO-ACN
66
to thinner, smoother fibres. The PEO-ACN solution has a high viscosity and low conductivity and
the ACN has a high vapour pressure. This suggests that the jet can dry fast before the filament
stretches significantly, resulting in wide fibres. The PEO-H2O-ACN solution has intermediate
conductivity, therefore intermediate stretching and evaporation. The fibre ‘bead on a string’
morphology suggests that surface evaporation has formed a solid surface shell while leaving a
liquid core that is still able to extend. These results demonstrate the dramatic effect on the fibre
morphology of the interplay of conductivity, viscosity and vapour pressure in electrospinning.
Electectrospinning polyethylene oxide with ammonia borane
It proved uncomplicated to dissolve varying amounts of AB into the 3wt% PEO (2M) in ACN
solution. The amount of AB added was such that the dry mass ratio of AB:PEO in the final
material would be 25:75, 50:50 and 75:25; these samples are termed the 25AB-75PEO, 50AB-
50PEO and 75AB-25PEO respectively. While the PEO solutions are highly viscous, it is shown
in Figure 30 that adding AB markedly reduces the viscosity.
Figure 30: The viscosity of the AB-PEO in ACN solutions with respect to shear speed. The viscosity at
the high shear limit quoted in Table 14. The addition of AB successively reduces the viscosity.
Table 14 lists the electrospinning parameters used when making the AB-PEO fibres. No Taylor
cone was observed but nonetheless the solutions could be stably electrospun. The 75AB-25PEO
solution was temperamental at times, probably due to the combination of the high conductivity
and low viscosity. The jet would break periodically, sometimes multiple jets could be observed
and once or twice solids built up on the nozzle. These issues would resolve themselves without
interference.
0 20 40 60 80 100 120 140 160 180 200 2200
1000
2000
3000
3% PEO (2M) in ACN
3% PEO (2M) in ACN with 25AB-75PEO
3% PEO (2M) in ACN with 50AB-50PEO
3% PEO (2M) in ACN with 75AB-25PEO
Vis
cosi
ty,
cP
Shear speed, RPM
Model Allometric1
Equation y = a*x^b
Reduced Chi-Sqr
4988.17588
Adj. R-Squar 0.9892
3% PEO (2M) in ACN
a
b
3% PEO (2M) in ACN with 25AB-75PEO
a
b
3% PEO (2M) in ACN with 50AB-50PEO
a
b
3% PEO (2M) in ACN with 75AB-25PEO
a
b
67
Table 14: Processing parameters used for the electrospinning the AB-PEO in ACN solutions.
AB:PEO dry
mass ratio
Flow rate
ml/hr
p.d.
keV
Conductivity
µSv/cm
Viscosity at
high shear, cP Observations
25:75 1 12 2.2±0.2 360 Stable spinning
50:50 1 12 7±2 350 Stable spinning
75:25 1 12 25±5 240 Temperamental spinning
The AB-PEO fibres are integral even in the 75AB-25PEO sample where the polymer is supporting
a large proportion of AB (see Figure 31). As the AB content increases, the fibres become rougher
and more uneven. While the fibres appear dry when electrospun at cross points in the 50AB-
50PEO and the 25AB-75PEO samples, the fibres have melded together. This suggests some
solvent has been retained in the fibres. To limit this residual ACN, the fibres were left open to the
air for 24 hours to dry.
Figure 31: Images of the AB-PEO fibre samples, 24AB-75PEO, 50AB-50PEO and 75AB-25PEO. As the
ratio of AB to PEO increases the fibres become rougher.
Uncertainty
In contrast to flat fibre mat obtained from PEO-H2O solution, the PEO-AB-ACN solutions
produced fluffy fibres (see Figure 32). To isolate the cause of this effect, PEO-AB-H2O solutions
were made and electrospun. The solution compositions and electrospinning parameters were kept
identical to those in Table 14 but the ACN was substituted for H2O. The fibres produced by the
PEO-AB-H2O solutions laid down flat. The ‘fluffy’ effect is probably due to an accumulation of
static charge. The ACN evaporates quickly and PEO is an electrical insulator so charge caught by
the AB would remain trapped.
As the pile of fibres builds up, the size of the whipping zone decreases. The initial fibres are
subjected to more stretching and drying time than the final fibres. This suggests variability across
the fibre sample.
25AB-75PEO 50AB-50PEO 75AB-25PEO
68
Figure 32: The fluffy fibres produced from the PEO-AB-ACN solution.
During the electrospinning of the PEO-AB-ACN solutions, bubbles evolved inside the syringe.
This suggests room temperature evolution of hydrogen from the AB in the solution. This is likely
as AB has been shown to decompose when refluxed with ACN [79].
This implies that both the nominal amount of pristine AB and the fibre production conditions are
subject to some ambiguity.
Freeze drying and dry mixing polyethylene oxide and ammonia borane
Electrospinning is an elegant but slow and labour intensive technique for producing AB-PEO
integrated materials and therefore two other methods, freeze drying and dry mixing, were
considered as alternatives. The freeze drying was relatively simple compared to electrospinning:
the solutions were prepared by dissolving the polymer in water and then AB was added to the
required dry mass ration AB:PEO of 25:75, 50:50 and 75:25. When the solids had blended with
the solvent, the mixture was placed in a freezer at -5°C and a day later transferred to the freeze
dryer. The dry mixing was done after grinding the solids with a pestle and mortar for ~15min.
X-ray diffraction of the electrospun and freeze dried polyethylene oxide and
ammonia borane composites
The XRD pattern of the PEO as-received granules and electrospun fibres display the monoclinic
(120) and (112) reflections at 19° and 23.2° respectively (Figure 33) [171], [172]. In the PEO
granules, numerous higher order peaks are also present and indicate a high level of crystallinity.
In the PEO fibres the peak intensity reduces and the higher order peaks disappear entirely. This
suggests the PEO fibres have less long range crystallinity than the unprocessed material. In the
solid state the PEO polymer backbone is twisted into a helix and these helices are laid parallel to
each other [173], [174]. The stretching forces experienced during electrospinning could distort
the polymer chain arrangement and limit the long range crystal ordering.
69
Figure 33: X-ray diffraction pattern of PEO as received granules and electrospun fibres.
In the electrospun composites, the tetragonal AB is clearly present in the 75AB-25PEO and
50AB-50PEO composite samples, as evidenced by the (110), (101), (200), (002), (121) and (112)
peaks in Figure 34 [84]. In the 25AB-75PEO sample the (110) and (101) peaks are reduced to a
small hump and the reflections at higher theta values are not visible either. This suggests that no
tetragonal crystalline AB is present in the 25AB-75PEO sample. The viscosity of the PEO
solution decreases with the addition of AB (Figure 30) indicating a high affinity between AB and
PEO. The data here implies that the low concentration of AB in the 25AB-75PEO solution inhibits
the formation of large AB crystallites, possibly because each AB molecule is more statistically
likely to interact with a PEO molecule then with itself. This suggests that the AB is so well
integrated with the PEO that the original crystal structure has been destroyed. This implies that
a different AB-PEO phase has been formed.
The (120) and (112) reflections from the PEO occur in some but not all of the composite fibres.
The quartet at 27° (marked +) is observed in all the composites but loses clarity when viewed
with the Philips diffractometer in the 50AB-50PEO and 75AB-25PEO samples. The quartet at
27° is associated with crystalline PEO. The changes in the (120) and (112) reflections between
the pure PEO and the PEO in the fibres suggests that monoclinic crystalline PEO is in some of
the fibres. The changes in the quartet at 27° suggest that monoclinic crystalline PEO is present in
all the fibres samples. To complicate the issue further new peaks have appeared at 10.6°, 12.1°,
13.4°, 14.9°, and 25.2° (marked *). These new peaks could be indicative of novel PEO phases or
novel AB-PEO phases. The novel peaks were observed in the 50AB-50PEO and the 75AB-
25PEO in the Rigaku but not in the Phillips diffractometer. Over the course of the study, fibres of
the same nominal AB and PEO contents were produced several times. The fibres tested on the
different instruments were from different batches. This suggests the preparation conditions can
70
effect the overal structure of the composites. Ultimately, it is clear that the AB and PEO have
interacted and formed a new phase with a well-defined crystal structure.
Figure 34: X-ray diffraction pattern of AB, PEO granules and the electrospun fibres collected on a
Phillips X’pert, above, and the Rigaku SmartLab, below.
Foam tests
One of the main barriers to commercialising AB is the foaming that accompanies the hydrogen
release. The foam occurs because the hydrogen gas is evolved inside the highly viscous liquid
that AB becomes on melting at 112°C. Figure 35 shows the typical response of a compressed
pellet of AB to heating; volume expansion of ~2000±140% is observed with a mass change of
15±8%. The volume change is probably greater than reported as the pellet rises up on the column
of foam until it is above the surface of the oil bath where the temperature is lower and so the
expansion ends.
Similar tests were performed with the electrospun fibres, freeze dried material and the hand milled
powder (Figure 36). In all examples, the foaming is successively supressed as the PEO content
increases. However, unlike for the case of pure AB, the composites all ejected white solids that
collected on the inside of the test-tube.
10 15 20 25 30 35 40 45 22 24 26
10 15 20 25 30 35 40 45 50 55 60
*
*
*
Phillips XRD
+
*
(112)
(120)
(110)(101)
(002)(112)
(121)
Inte
nsi
ty2
PEO powder
25AB-75PEO
50AB-50PEO
75AB-25PEO
AB powder
(220)
*
}
**
*
Rigaku XRD
+
*
* **
(120)
(112)
(022)
(220)
(130)(013)(112)(002)
(121)(200)
(101)
Inte
nsi
ty
2
(110) Instrument
}
71
Figure 35: Foaming response of a compressed pellet of AB when subjected to 120°C isothermal heating
For the high ratio AB samples (75AB-25PEO) the electrospun fibres outperformed the freeze
dried and hand milled materials. On lowering the AB content (50AB-50PEO) the hand milled
material had the best antifoaming properties as the pellet shrank, but the freeze dried material
ejected fewer solids. The low AB content materials (25AB-75PEO) in general expanded more
than their 50AB-50PEO counterparts. The electrospun 25AB-75PEO sample expanded vertically,
the freeze dried 25AB-75PEO sample unravelled and the hand milled 25AB-75PEO sample
expanded equally in all directions. This suggests that for 25AB-75PEO samples have expanded
along the packing axis. The freeze dried and electrospun pellets are made by folding the original
sheet like material into the pellet press. Each fold is held under tension from the compression but
the expanding gas seems to give it power to unfurl (Figure 37).
The large error bars in Figure 36 suggested a factor other than the AB content is affecting the
foaming of the composites. When a large number of pellets from a single sample were foam
tested, a positive trend was observed between the packed density of the pellet and the expansion
in the foam test (Figure 36). With a higher density there will be fewer cavities and gas release
pathways in the pellet. This suggests that the gas escape will be slowed and a greater pressure will
be created in higher density pellets. More compressed gas will eventually form larger bubbles and
the overall pellet expansion will be greater.
The 50AB-50PEO samples are unusual in that they seem to shrink as often as expand. When the
pellets are closely studied, they expand in the vertical axis but the cross section shrinks, overall
resulting in a final reduction in volume. At 120°C melting of the AB and PEO and foaming of the
AB are occurring. It is likely, that the foaming is causing the vertical expansion while the AB and
PEO are melting into the free volume in the initial pellet.
Clear
Foam
Unexpanded head
Oil Bath
120°C
72
Figure 36: Foam tests with pellets made from electrospun fibres, freeze dried material, and hand milled
powder with AB contents, 25, 50 and 75wt%
25AB-75PEO 50AB-50PEO 75AB-25PEO
Electrospun
Freeze dried
Hand milled
73
Figure 37: Unfurled pellet
Figure 38: Representative examples of the relation between initial pellet density and foaming.
The final consideration is the white residue that appears on the inside of the test-tube ejected from
the AB-PEO samples. When the pellets in the test-tube are immersed in the hot oil, a period of
inactivity (about 1 minute long) precedes the foaming. The foam is accompanied by particulates
some which seem to explode off the pellet and some which travel out of the test tube in the gas
stream; both types collect on the inside of the glass. If an AB composite of this type is to be
commercialised particulate filters would be necessary to keep solids from poisoning the fuel cell.
On the whole PEO has a beneficial effect by acting as an antifoaming agent and is a suitable
additive to AB. With low AB content, 50% and less, the pellet made from the compressed hand
milled material foams the least but when the AB content is raised to 75% the electrospun material
has the best characteristics.
0.5 0.6 0.7 0.8 0.9 1.0
-50
0
50
100
150
200
250
300
350
400
450
Vo
lum
e C
han
ge,
%
Initial Density, g/cc
50AB-50PEO Hand milled
75AB-25PEO Electrospun
Pellet press
Unfurled pellet
Mat like freeze dried 25AB-75PEO material
74
Ramp heating of the ammonia borane-polyethylene oxide composites and
pristine ammonia borane
The AB-PEO composites were analysed in a combined thermogravimetric analyser (TGA) and
differential scanning calorimeter (DSC) with attached residual gas analyser (RGA). The samples
were subjected to ramp heating runs of 2°C/min, between room temperature and 200°C and
isothermal heating runs at 80°C, 100°C, 120°C and 150°C. The RGA data are normalised to the
mass of AB in the sample and to the argon flow, the TGA data are normalised to the nominal AB
content and the DSC data are normalised to the total sample mass.
Hydrogen release, thermogravimetric analysis and differential scanning
calorimetry
In Figure 39 the RGA, TGA and DSC of the pristine AB and the AB-PEO electrospun samples
are plotted. The TGA hydrogen (m/z=2) trace for pure AB has a profile that is typically associated
with pure AB [55], [56]. The hydrogen is released in two stages, the first peaking at 110°C and
the second at 150°C. The two stages are the release of the 1st and 2nd hydrogen equivalents (H2
equivs.). In the TGA, the AB mass loss curve is steepest when the hydrogen signal is highest,
implying hydrogen liberation is responsible for the change in sample mass. However, the AB has
lost approximately 50% of its mass and as the 2 H2 equivs. account for just 13wt% of AB, other
entities besides hydrogen are being released. In the DSC curve an endothermic melt dip,
beginning near 95°C with a minimum at 105°C, occurs before the hydrogen is released. The main
melting peak is accompanied by a small pre-dip. This is possibly a signal of eutectic impurities
in the raw AB [175]. The AB used for the experiments is 97% pure and the impurities present
could perhaps influence areas of the solid to have lower melting points. Or alternatively this
feature is the result of the overlay of the exothermic hydrogen release and the endothermic
melting. The melting of AB is the result of the dihydrogen bonding network breaking; the AB
molecules become mobile and can react easily [176].
The pure PEO does not release hydrogen (not shown) and does not show any mass change over
the course of the heating ramp (TGA curve). The DSC trace shows a melting endotherm which
begins at 60°C and reaches the minimum at 69°C, this is comparable to the literature melting
temperature of PEO (67°C) [103].
Adding PEO to AB dramatically changes the H2 release profile (RGA), the mass loss (TGA) and
the heat flow (DSC). Increasing the PEO content in the fibres (75AB-25PEO50AB-
50PEO25AB-75PEO) successively lowers the temperature of the onset of the release of the 1st
and 2nd H2 equivs. The associated mass loss also occurs at lower temperatures and reaches a
maximum of between 45% and 55% of the initial AB mass. Like with the pristine AB, the
75
approximately 50wt% mass loss is too high to be due only to hydrogen. In the DSC, the
exothermic peaks for the 1st and 2nd H2 equivs. also occur at lower temperatures as the PEO content
increases. This implies that the PEO is actively encouraging the AB to react at lower temperatures.
Figure 39: Combined TGA, DSC and RGA hydrogen signal (m/z=2) for pristine AB and the AB-PEO
electrospun fibres. As the fraction of PEO in the sample increases, the hydrogen release is activated at
earlier temperatures. Mass loss occurs earlier in the AB-PEO composites than in the pristine AB but the
final wt% loss is similar (~50%) in both. In the DSC curves, successive addition of PEO reduces the
depth of the AB melting endotherm. The lower temperature release of hydrogen is accompanied by an
exotherm which reduces the apparent size of the melting endotherm.
There are a few interesting things to notice in the DSC trace in Figure 39. In the composites the
PEO melting dip minimum moves successively upwards in temperature (74°C to 76.5°C to
0.4
0.6
0.8
1.0
60 80 100 120 140 160 180-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Temperature, oC
Temperature, oC
no
rmal
ised
to
mas
s o
f A
B a
nd
arg
on
flo
w,
a.u
.
PEO powder
Pure AB
75AB-25PEO
50AB-50PEO
25AB-75PEO
0
5
10
15
60 80 100 120 140 160 180
DS
C,
mW
/mg
PE
O,
no
rmal
ised
to
mas
s o
f P
EO
no
rmal
ised
to
no
min
al m
ass
of
AB
in
th
e sa
mp
leM
ass
loss
,
Hy
dro
gen
sig
nal
, m
/z =
2,
76
78.5°C) as the AB content decreases from (75AB-25PEO50AB-50PEO25AB-75PEO).
Also, as expected, the area of the dip reduces as the PEO content is reduced. The electrospinning
process is known to align and increase the crystallinity of polymers [129]. Greater crystallinity in
a polymeric material will push the melting point upwards. This could account for the rise in
melting temperature. Alternatively, intermolecular hydrogen bonding between the PEO and AB
could be occurring. These bonds would likely be stronger than the PEO to PEO inter-chain
interactions because of the greater polarity/dielectric constant of AB, requiring more heat energy
to melt them.
The AB melting curve initiates at lower temperatures as the PEO content in the fibres increases.
Further, the depth of the AB melting curve, decreases as a fraction of the height of the following
exothermic peak, as the PEO content increase. This suggests two things. Firstly, the PEO is
lowering the melting point of AB, perhaps through the AB to PEO bonds suggested above.
Secondly, the exothermic peak due to the release of the 1st H2 equiv. is also occurring at lower
temperatures. It is possible that the onset temperature for the exothermic peak has reduced a
greater degree than the onset temperature for the AB melting. As the melting has not completely
disappeared in the samples, it is likely that some of the AB is in its bulk state. The PEO and AB
melting peaks are well separated so it is unlikely that serious AB melting is occurring at the same
time as the PEO.
In the X-ray diffraction pattern the 75AB-25PEO and 50AB-50PEO fibres show evidence of
containing tetragonal AB by the intense (110) and (101) reflections. In the 25AB-75PEO sample
these peaks have all but disappeared so the AB must exist in a different form.
It can be concluded that the AB and PEO have combined into a unique, new phase. It is this phase
that initiates the reaction at lower temperatures without requiring the AB to melt.
The freeze dried materials were also tested under similar ramp heating runs. The results obtained
are indistinguishable from the electrospun materials.
Impurities
Cumulative integrals of the heating ramps of pure AB and the AB-PEO composite materials were
compared to study the impurity levels in the gas stream. For reasons explained in chapter 3, the
RGA signal and the normalising technique are not adequate to provide a robust measure of the
differences between the quantity of hydrogen released from pure AB and the AB-PEO
composites. However, it is possible to compare between the plots for AB-PEO samples.
77
The hydrogen (m/z=2) release in Figure 40 shows as before, that the PEO reduces the temperature
of the onset of hydrogen release. Further, with more PEO the slope of the graph steepens. In the
pristine AB, the steps associated with the release of the 1st and 2nd H2 equiv. are clearly
differentiated. As the PEO fraction in the fibres is raised (75AB-25PEO50AB-50PEO25AB-
75PEO) the release of the 2nd H2 equiv. moves down and overlaps the peak for the release of the
1st H2 equiv.
In the composite materials the levels of borazine (m/z = 81) and diborane/borazine fragments (m/z
= 27) as a fraction of the hydrogen signal, increase as the PEO fraction increases. Additionally,
borazine and diborane first appear from the composites at 105°C, a much lower temperature than
the pristine AB at around 145°C. Borazine, B3N3H6, has one hydrogen atom per base atom (B or
N) and is only formed during the release of the 2nd H2 equiv. [46], [53], [59]. The earlier release
of boron-containing impurities is consistent with the drop in temperature of the release of 2nd H2
equiv. already observed.
In Figure 40 the levels of water (m/z = 18) and ammonia (m/z = 17), remain similar in the three
AB-PEO composites. Water is the principal impurity in the gas stream, and accounts for around
10% of the gas detected in the RGA. As water is nine times heavier than hydrogen this could go
some way to account for the massive and unexpected mass loss in Figure 39. Water has not been
used in electrospinning the composites but AB has probably absorbed moisture from the air [177].
Acetonitrile was the solvent used in the electrospinning and some residue is expected in the fibres,
however it was not observed in the RGA.
Conclusions
From this data several things can be concluded about the effect of PEO on AB. Firstly, PEO is
helpful in nucleating the dehydrogenation of AB. Additionally, it encourages all stages of the
reaction to occur at lower temperatures. Borazine is observed in the AB trace, but only at high
temperatures during the release of the 2nd H2 equiv. When PEO is added, the ratio of borazine to
hydrogen is increased. This could be a feature of the reaction being further along at lower
temperatures, or perhaps the PEO is encouraging the formation of cyclic products. Water might
be responsible for some of the excessive mass observed.
Isothermal heating
Hydrogen release under isothermal heating runs
In addition to the heating ramps the AB-PEO fibres and freeze dried composites were subjected
to isothermal heating runs in the TGA/DSC with attached RGA. Isothermal runs at 80°C, 100°C,
78
120°C and 150°C were performed. The RGA data were normalised to the nominal AB content
and the argon level.
Figure 40: Cumulative integrals of the heating ramps of pristine AB and the AB-PEO composite fibres. It
is noted that raising the amount of PEO in the mix aids the hydrogen release but produces more borazine.
The data were normalised to the hydrogen (m/z = 2) level at 190ºC and errors in the data are shown as the
lighter background.
79
Following the hydrogen (m/z = 2) trace in Figure 41 demonstrates some of the properties of the
composites and AB. Firstly, at a constant 80ºC pristine AB evolves minimal hydrogen. When
PEO is added (75AB-25PEO sample), hydrogen is produced and if the PEO content in the
composite is raised (25AB-75PEO sample), hydrogen is released at earlier times. In fact, at all
temperatures, as the proportion of PEO is increased the onset and the peak of the main hydrogen
release will tend to earlier times.
At 80ºC and indeed the other higher temperature runs, a pre-peak (marked *) is visible before the
main hydrogen release peak. This pre-peak is only observed in the composites not in the pure AB.
Also, the pre-peak decreased in size (with respect to the main hydrogen release peak) as the
temperature was raised. This suggests the earlier occurrence of the main hydrogen release peak is
over shadowing the pre-peak at 150 ºC. This pre-peak is likely the hydrogen release from the new
AB or AB-PEO phase observed in the XRD pattern (Figure 34).
The sample temperature lags behind the set temperature, only achieving its maximum temperature
between minutes 10 and 20. This and the ramp heating data indicates that rapid hydrogen release
occurs above 80ºC in the AB-PEO fibres without the preceding induction period observed for
AB.
The shape of the traces is similar for the AB, 75AB-25PEO and 50AB-50PEO samples. The
25AB-75PEO sample is much broader. Possibly more hydrogen is being released overall (1st and
2nd H2 equivs.) from the 25AB-75PEO sample but this has not been verified.
Non-hydrogen species released form AB under step heating
Fuel cells require a stream of pure hydrogen to work efficiently and contaminants in this gas
stream will poison the cell and reduce its lifetime. For this reason additional species besides
hydrogen that are released during the dehydrogenation of AB are undesirable and should be
controlled as far as possible to minimise the use of filters in the system.
The presence of borazine in the gas stream is influenced both by temperature and the PEO fraction
in the material. In Figure 42 at 80°C no borazine is observed. At 100°C, only the 25AB-75PEO
sample releases borazine, the sample with the greatest PEO content. At 120°C all the AB-PEO
composites release some borazine but the pristine AB does not. At 150°C, the borazine released
by the AB-PEO has increased while the pristine AB appears not to release any. In the ramp heating
runs, pure AB began to release borazine at about 145°C so we should expect to see it here. Closer
examination of the raw data shows that while the furnace is set for a particular temperature, the
sample temperature is recorded between 4°C and 6°C lower: the 150°C heating run is actually at
146.5°C (see Figure 23). Borazine is released as a by-product during in the release of the 2nd H2
80
equiv., as this begins above 150°C [55], [56] we would not expect to detect borazine in the gas
stream from pristine AB.
The borazine is release is coordinated with the release of the 2nd H2 equiv. And the ramp heating
runs showed that the PEO accelerated the dehydrogenation so that the release of the 2nd H2 equiv.
occurs at lower temperatures. The increased borazine detected from the AB-PEO composites
could be due to the accelerated hydrogen release. Or the PEO could be actively catalysing the
formation of borazine.
The diborane release, in Figure 43 is similar to that of borazine. At low temperatures, 80°C and
100°C no diborane is observed, while at high temperatures, 120°C and 150°C more borazine is
observed form the AB-PEO fibres then for the PEO.
As in the heat ramps, the ammonia signal in Figure 44 in the isothermal heating steps remains
fairly constant sample to sample. At 80°C the pure AB does not release ammonia, this is however
unsurprising considering no hydrogen is released either. At 100°C and above the pure AB does
release ammonia. The PEO-AB samples all release ammonia, and the release occurs earlier as the
temperature is raised like the hydrogen.
Conclusions
The dehydrogenation of AB under isothermal conditions initiates at earlier times as the PEO
content is raised. However, increased borazine levels are also observed.
81
Figure 41: Hydrogen (m/z=2) trace of the isothermal heating runs of pure AB and the composite
materials. At all temperatures, the dehydrogenation of AB begins earlier and is accelerated as the PEO
fraction of the composite increases.
20 40 60 80 100 120 140 160
0.000
0.015
0.030
0.045
0.0
0.1
0.2
0.0
0.5
1.0
10 15 20 25 30 35 40 45 500
1
21500C
1200C
1000C
PureAB
75AB-25PEO
50AB-50PEO
25AB-75PEO
800C
*
*
*
Time, min
Hyd
rogen
sig
nal
, m
/z =
2,
norm
alis
ed t
o m
ass
of
AB
an
d a
rgon
flo
w,
a.u.
Time, min
82
Figure 42: The borazine signal, m/z=81, from pristine AB and the AB-PEO fibres under isothermal
heating at 80°C, 100°C, 120°C and 150°C. The pristine AB does not release any borazine at any of the
temperatures. This is consistent with literature and the heat ramp data (Figure 40) where borazine is first
observed at 150°C. (The sample is actually several degrees cooler than the written value, see
experimental chapter 3). None of the AB-PEO fibres release borazine at 80°C but all give off borazine at
120°C and 150°C. At 100°C the 25AB-75PEO is the only sample that releases significant borazine. This
is consistent with the ramp heating tests where it is first observed around 100°C in this sample. This
suggests the PEO is encouraging borazine to be released.
10 20 30 40 50 60 70 80 90 100 110 120
0.000
0.001
0.002
0.000
0.001
0.002
0.003
0.000
0.003
0.006
10 15 20 25 30 35 40 45 500.00
0.02
0.041500C
1200C
1000C
PureAB
75AB-25PEO
50AB-50PEO
25AB-75PEO
800C
Time, min
Hydro
gen
sig
nal
, m
/z =
2,
norm
alis
ed t
o m
ass
of
AB
and a
rgon f
low
, a.
u.
Time, min
Borazine
83
Figure 43: The diborane signal, m/z = 27, released by pure AB and the AB-PEO fibres under isothermal
heating of 80°C, 100°C, 120°C and 150°C. No diborane was detected at 80°C and 100°C. At 120°C both
the pristine AB and the AB-PEO fibres show borazine. At 150°C the diborane level are higher in the AB-
PEO fibres than in the AB. This suggests PEO encourages the formation of diborane at high temperatures.
20 40 60 80 100 120 140 160 180 200
0.000
0.001
0.002
0.000
0.005
0.010
0.000
0.003
0.006
10 20 30 40 500.0
0.2
0.41500C
1200C
1000C
PureAB
75AB-25PEO
50AB-50PEO
25AB-75PEO
800C
Time, min
Hy
dro
gen
sig
nal
, m
/z =
2,
no
rmal
ised
to
mas
s o
f A
B a
nd
arg
on
flo
w,
a.u
.
Time, min
84
Figure 44: The ammonia signal, m/z= 17, released by pristine AB and the AB-PEO fibres under
isothermal heating of 80°C, 100°C, 120°C and 150°C. The ammonia levels are similar in the pure AB and
the fibres, but ammonia is released sooner in the fibres. At 80°C AB does not release any ammonia, but
then it doesn’t release any hydrogen either. This suggests PEO does have a significant effect on the
ammonia release.
20 40 60 80 100 120 140 160
0.000
0.005
0.00
0.02
0.04
0.00
0.02
10 15 20 25 30 35 40 45 500.0
0.1
0.2
1500C
1200C
1000C
PureAB
75AB-25PEO
50AB-50PEO
25AB-75PEO
800C
Time, min
Am
monia
sig
nal
, m
/z =
17,
norm
alis
ed t
o m
ass
of
AB
and a
rgon f
low
, a.
u.
Time, min
85
Fourier transform infrared spectroscopy
Fourier transform infrared spectroscopy was performed on the as-received AB and the as-
electrospun PEO and AB-PEO fibres to study changes that may have occurred in either the AB
or the PEO structure and intermolecular interactions as a result of being in a combined composite.
Further, samples of the 50AB-50PEO fibres have been heated at 85°C ex-situ for different periods
of time (up to 2 hours) to examine how the bonding is affected as the hydrogen is released. The
absorption peaks of the pristine AB sample were referenced from Zhang et al. 2011, Demirci et
al. 2011 and Xie et al. 2009 [178]–[180] and absorption peaks of the PEO have been assigned to
Gondaliya et al. 2011 and Sundar et al 2006 [181], [182].
The experiment being performed on the solids ex situ, ammonia, diborane, borazine and other
liquid or gaseous products form AB are not likely to be observed.
FTIR of ammonia borane and polyethylene oxide
The AB and PEO FTIR spectra in
Figure 45 are typical of literature, the N-H, B-H and C-H stretching modes are well represented
in the high wavenumber end of the spectrum and even the peaks in the fingerprint region are
distinct and were easily assigned [178]–[182]. The N-H and B-H stretching modes are wide and
diversified as they are a collective of the asymmetric and symmetric stretches and may include
NH4 and BH4 moieties.
The intensity of a peak in the infrared spectrum is determined by the magnitude of the change in
dipole moment induced by the vibration [152]. Because of this, infrared spectroscopy cannot be
used as quantitative tool except when dealing with peaks of identical origin. Because of the greater
difference in electronegativity the N-H bands are expected to be of higher relative intensity than
either the C-H or B-H bands. However in our samples, rather surprisingly, the spectrum is
dominated by the BH bands and this may be a direct reflection of the strong N->B dative
interaction which affects the charge at the boron centre.
The electrospun PEO is crystalline as defined by the triplet peaks 1147cm-1, 1097cm-1 and
1060cm-1 [89], [181].
86
Figure 45: The FTIR spectrum of as received AB, electrospun PEO fibres and 50AB-50PEO fibres.
Stretches are identified from literature. [178]–[182].
FTIR of the AB-PEO fibres
The AB-PEO electrospun fibres clearly display features from AB and PEO as well as some
additional stretches (Figure 46). This suggests a combination of areas of bulk AB and PEO and
also mixed regions where new interactions between the AB and PEO can exist.
A change in the hydrogen bonding will alter the peak position because the strength of the
hydrogen bond inversely affects the length and therefore vibrational frequency of the covalent
bond [151], [152], [183]. In a simple case, weakening the hydrogen bonds between the AB
molecules will move the N-H and B-H stretches to a higher wavenumbers. In Figure 46 in the
fingerprint region of the AB-PEO samples, peaks corresponding to the NH3 and BH3 have
undergone blue shifting (highlighted in grey). The NH3 symmetric and asymmetric deformations
at respectively 1599cm-1 and 1376 cm-1 both produce peaks at higher wavenumber (1635 cm-1 and
1401 cm-1 respectively). The BH3 deformation (B-H scissor) at 1154cm-1 and the N-B-H rocking
3500 3000 2500 1600 1400 1200 1000 800 600 400
0.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
Wavenumber, cm-1
Wavenumber, cm-1
asymmetric
H-wagging
symmetric
asymmetric
Wavenumber, cm-1
NBH rocking
asymmetric
symmetric B-H stretch
Abso
rban
ce, %
AB
N-H stretch
}
N-B stretch
H-wagging
HB3 scissor deformation
NH3 scissor deformation
symmetric
3500 3000 2500 1600 1400 1200 1000 800 600 4000.0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
0.6
0.8
PEO
and CH2 rocking
C-O-C bending
CH2 rocking
C-O-C vibration
C-O stretch
CH2 twist
CH2 wagging
CH2 scissor
C-H stretch
87
at 1056cm-1, display a new shoulder in the AB-PEO samples at 1165cm-1 and 1077cm-1
respectively.
In addition to the differences observed in the FTIR spectra between the pristine AB and PEO and
the AB-PEO samples, there is variation between the three AB-PEO samples. In Figure 46, moving
from the 75AB-25PEO through to the 25AB-75PEO sample, a successive increase in peak
intensity occurs in the new bands at 1635, 1402, 1187, 1165 and 1077cm-1 (grey highlights). In
the high frequency region a similar shift is faintly observed (grey highlights). This, and the other
peaks highlighted in grey shows that the blue shift of the AB peaks is enhanced with the increased
availability of PEO.
Correspondingly, the PEO peaks at 1096 cm-1 (C-O stretch) and 1279 cm-1 (CH2 twist) have both
split into two overlapping peaks in the fibres at 1105 and 1093 cm-1 and 1285 and 1275 cm-1
respectively (dashed lines).
These bond vibrational changes are expected due to the crystal structure changes of the AB and
PEO previously observed in the XRD pattern (Error! Reference source not found.). Across the
fibre samples, the intensity of the new bands scale with the PEO content, yet the peak positions
are constant. This shows that the new features present in the fibres represent an ideal, well-defined
and preferred arrangement between AB and PEO. The increase in bond vibrational frequency
observed indicates a reduction of AB intramolecular bond lengths which correspond to a
lengthening and weakening of intermolecular bonds - the dihydrogen bonds.
Here it is conceivable that the self-to-self dihydrogen bonding is partially replaced by AB-to-PEO
hydrogen bonding. Possibly electron donation from the oxygen lone pair to the AB has increased
the electron density on the N-H and B-H bonds which shortens them [152].
Isothermal heating of 50AB-50PEO studied by FTIR
In Figure 47 in the high frequency part of the spectrum, 4000 to 2000cm-1 over two hours at 85°C
the N-H and B-H stretching regions lose intensity while the C-H stretching peak remains constant.
Clearly hydrogen loss is only from the AB and occurs at both the nitrogen and boron ends of the
molecule. The peaks corresponding to the PEO in the fingerprint region also retain their intensity
throughout the experiment, while those corresponding to N-H and B-H diminish.
88
Figure 46: The FTIR spectra of AB, PEO and the three electrospun fibres. As the PEO content in the
fibres increases there is a change in intensity (grey highlights), a shift to higher wave number.
Additionally there is splitting of the PEO C-O and C-H stretching peaks (dashed line). This supports the
previous suggestion that the dihydrogen bonding of AB is being replaced by hydrogen bonding between
AB and PEO.
Over the course of the experiment several new peaks arise; they are concentrated in the fingerprint
region and their positions are marked in Figure 47. Of these, a significant proportion have been
identified as originating from a boron to oxygen (B-O) bond. Considering the proximity of the
AB and the PEO in the electrospun fibres, an interaction between the electron donating oxygen
and the electron accepting boron is highly likely and these peaks in Table 15 substantiate this
hypothesis.
1700 1600 1500 1400 1300 1200 1100 1000 900 800 700
AB
PEO
Wavenumber, cm-1
75AB-25PEO
50AB-50PEO
25AB-75PEO
Abso
rban
ce
3500 3000 2500 2000
AB
PEO
Wavenumber, cm-1
75AB-25PEO
50AB-50PEO
25AB-75PEO
Ab
sorb
ance
89
Table 15: The stretching frequencies observed in the FTIR that correspond to B-O bonds
Wavenumber, cm-1 Description [87], [184], [185]
1402 B-O Asymmetric stretch
1024 B-O Interaction
807 & 887-872 & 920-925 B-O Symmetric stretch
696 & 665 B-O out of plane bend
Previously, the peak at 1401cm-1 has been associated to an overall weakening of the hydrogen
bonding of the NH3 group and to the symmetric stretching of a newly formed B-O bond. There is
third contender for the peak which is a group NH4+ noted by Demirci et al [179] who put the B-
O asymmetric stretch at 1450cm-1 [186].
If an NH4+ group were present, IR bands would occur also at 3145cm-1 (stretching), 3040cm-1 (N-
H stretch) and 1680cm-1 (N-H deformation) as well as the 1402cm-1 (bending) [187]. None of
these additional bands were observed suggesting that no NH4+ was present. Thus the band at
1402cm-1 is either a B-O stretch or a change in the dihydrogen bonding of the AB NH3 group. It
is most likely that it is due to a B-O stretch as it was only observed in the heated sample.
The original B-N bending peaks at 726cm-1 and 783cm-1 (B-N stretch) disappear on heating and
simultaneously two peaks at 753 cm-1 and 758 cm-1 appear. Taylor and Cluff observed a shift to
lower wavenumber of the B-N band as the hydrogen atoms were replaced with deuterium [188].
The two new peaks are likely the result of a change in the B-N stretch, either DADB, linear dimers
or PAB. This suggests they are related to B-N bonding, probably DADB, PAB, or other fragments
and cyclic products associated with PAB. The two new peaks at 479 cm-1 or 465 cm-1 could not
be identified.
Conclusions
These data suggests the PEO and AB interact via hydrogen bonding in the 50AB-50PEO
electrospun fibres. On heating hydrogen, loss occurs at both the nitrogen and boron side of AB.
Boron to oxygen bonds form. There is no oxygen in the pristine AB. This suggests that the boron
to oxygen reaction could be part of the reason the AB in the fibres decomposes at an accelerated
rate and produce borazine.
90
Figure 47: FTIR of the heated samples. New peaks that appear during the heating are labelled.
91
Nuclear magnetic resonance spectroscopy
11B NMR of ammonia borane heated in situ at 85°C
NMR spectra were initially recorded at room-temperature before the temperature was rapidly
increased to 85°C, where it was kept constant for the duration of the experiment. In situ 11B NMR
spectroscopy was performed on a pristine AB sample and on the 50AB-50PEO electrospun
material. Any gaseous products were able to evaporate during the experiment so we do not expect
to observe any peaks associated with ammonia, borazine or diborane.
Figure 48: 11B NMR of pure AB at 85°C.
Assignment of NMR spectra peaks in Table 16 was made with reference to literature including
both experimental [66], [78] and theoretical studies [82] of AB and its reaction products.
-40 -30 -20 -10 0
BH2 (linear dimer)
AB (BH3)
ppm
Minutes
0
55
102
155
208
314
349
BH4 (DADB)
BH3 (AB*, mobile phase)
BH3 (end group)
BH2 (DADB)
BH groups
92
Table 16: NMR peak assignments.
Signal (ppm) Group
-38 BH4 (DADB) [66]
-25 & -28 BH3 (AB) [66]
-23 BH3 (AB*) [66] or end group [82]
-21.7 BH3 (end group, linear dimer) [66]
-13.5 BH2 (DADB) [66]
-12.5 BH2 [66] (linear dimer) [66]
-11.2 BH2 [66] (PAB) [82]
-9.9 BH2 (PAB) [82]
-7 BH (PIB or branched PAB) [82]
1.2 BO interaction [87]
-14 BO interaction [87]
In Figure 48, BH3 peaks corresponding to AB were initially observed at -25ppm and -28ppm,
these signals reduce in intensity as a signal at -23 ppm corresponding to the mobile phase of AB
(AB*) increase. (AB* is not the melting of AB [66].) As the AB* peak drops, signals
corresponding to the diammoniate of diborane (DADB) at -38ppm (BH4) and -12ppm (BH2) grow
[66], [78]. Shortly after, the combined very broad signals at around -5 ppm, assigned to BH and
the peaks at -12 ppm, belonging to the BH2, suggest the presence of either a branched or cyclic
PAB or PIB. The spectrum at 208 minutes has high resolution in the BH2 region. Distinct peaks
are observed at -13.5, -12.5, -11.2 and -9.9ppm that are indicative of discrete BH2 environments.
These peaks are consistent with the presence of DADB (BH4 at -38ppm and BH2 at -12ppm) [66],
linear dimers (NH3BH2NH2BH3) (terminal BH3 at -21.7ppm and BH2 -12.5 or -11.2ppm) [66] and
tentatively, PAB (BH2 -11.2 or -9.9ppm) [82].The simultaneous presence of several species is
reasonable given the long timescale of the reactions and is attributed to the reaction being at
different stages in different areas of the sample.
On close observation the peak assigned to the BH4 group appears to be the sum of two peaks (-
37.5 and -38.2ppm). It is suggested that this difference is due to the existence of two distinct BH4
environments. At -38.2ppm the BH4 is associated with DADB, the shift to -37.5ppm indicates
decreased symmetry about the boron and such as might be experienced in the vicinity of PAB
moieties [189]. This is supporting of the nucleation and growth mechanism proposed by Shaw et
al [78] where PAB chains are constantly lengthened by -BH4 working to add other AB molecules
but is never used up itself. At 314 minutes it is suggested that the dominant species switches from
DADB to PAB, thus the peak assigned to BH4 shifts from -38.2ppm to -37.5ppm. By the end of
the experiment, at 349 minutes, the progress of the decomposition is indicated by the broad peak
at -7 ppm, which is assigned to BH and indicates the presence of PIB. It is difficult to produce
93
PIB from AB at 85°C, however, as the pristine AB has been heated for 6 hours it is conceivable
that in some areas of the sample more than 1 equiv. of hydrogen has been lost.
11B NMR of the 50AB-50PEO fibres heated in situ at 85°C
Figure 49: 11B NMR of 50AB-50PEO material heated at 85°C for several hours.
In the 50AB-50PEO sample, the spectra are similar in many ways, though dehydrogenation occurs
significantly faster – AB* is observed within 4 minutes and DADB within 8. Unlike the AB
sample, the BH2 area is not well resolved, nor is there any evidence for a terminal BH3 group.
This is suggestive of the formation of a profusion of cyclic products [78], precursors to the
borazine observed in the RGA.
The BH4 peak shifts from -36.7 to -38.3ppm over the course of the dehydrogenation suggesting
two distinct environments for the BH4+ ion. These are assigned to DADB and PAB respectively,
and unlike the AB sample, are observed to coexist for much of the experiment. The coexistence
of these environments may be a result of the greater reaction speed. This would allow at time 33
minutes of heating, where in some regions of the sample the reaction has already progressed to
the end, while in other parts, DADB is still being nucleated.
Another difference is that as the AB* peak fades (55 minutes onwards) while intensity at -25ppm
(also corresponding to a BH3) emerges. Possibly with continued heating, the pure AB will
-40 -30 -20 -10 0 10
B-O interaction
BHBH
2 groups
BH3 (AB*, mobile phase)
-BH4 DADB
AB
ppm
Minutes
0
4
8
11
19
22
33
55
102
261
94
eventually demonstrate the same peak, or this could be an artefact of different reaction products,
maybe (cyclodiborazanyl) aminoborohydride (BCDB) as this would fit with the shift to lower
energy of the BH2 peaks [78],
Additional peaks were observed at 1.2 and 14ppm which suggests the presence of a B-O bond
[87] Chen et al also observed a peak at -0.9 ppm from the interaction of BH3 with the oxygen on
THF [125]. This could form as the result of a lone pair donation from the oxygen atom (in the
PEO chain) to the boron atom (in AB) or one of its decomposition intermediates.
Conclusions
In Figure 50 the spectra of the AB and 50AB-50PEO have been superimposed to clearly show the
differences. It is proposed that during electrospinning, an intimate mix of PEO and AB is formed.
The oxygen in PEO has a partial negative charge, Oδ-, and the ability to hydrogen bond via lone
pair donation. This composite should therefore contain hydrogen bonds between AB and PEO, as
well as the traditional dihydrogen bonds between AB molecules. In the 50AB-50PEO sample, the
AB* peak is present before any heating has occurred. The formation of AB* being a rate limiting
step in the AB dehydrogenation, this could explain the increased reaction kinetics of 50AB-
50PEO material. Solid AB and AB* seem to coexist for longer in the 50AB-50PEO, likely the
result of the far faster reaction time in the fibres than in the case of pristine AB. In the 50AB-
50PEO the B-O bonds appear at the same time as the BH4 and BH2 groups. This suggests that the
forming of B-O bonds could be the activator for the early appearance of DADB. The width and
position change of the BH4 peak is discussed in the final conclusions.
95
Figure 50: The 11B NMR studies of AB dehydrogenation at 85°C. The pristine AB superimposed on the
50AB-50PEO fibre sample. The AB in the fibres reacts faster, the BH4 group has a broader peak and B-O
bonds occur.
15N NMR
The natural abundance of 15N is very low, <1%, however it is still possible to collect NMR data
if the cycle is run for long enough. Data were collected (at room temperature) from the pristine
AB and the as-spun fibres and AB and fibres that had been heated in an oven at 85°C for 6 hours.
In Figure 51 the pristine AB and the AB in the fibres both demonstrate a strong peak at -366.8ppm
-40 -39 -38 -37 -36 -32 -30 -28 -26 -24 -22 -10 0 10 20
Minutes
Pristine AB
0
55
102
155
208
314
349
50AB-50PEO
0
4
11
19
22
33
102
261
ppm
96
indicating the presence of tertiary amines, NH3. However, two small extra peaks (-371.3ppm and
-374ppm) are observed for the AB in the 50AB-50PEO. This indicates that the nitrogen
environment, for a portion of the AB molecules, changed when PEO was added to the mixture. A
shift down field of the nitrogen peak can be due to increased shielding from the magnetic field
due to increased electron density on the nitrogen. As suggested above, the most likely explanation
is that a portion of the N-H groups are hydrogen bonding to the oxygen in preference to B-H
groups.
Figure 51: 15N NMR of pure AB and 50AB-50PEO electrospun sample before and after 6 hours heating
85°C when PAB is observed.
The majority of the pure AB transforms to PAB [190] after 6 hours of heating at 85°C. This is
expected as in six hours at 85°C AB can release one equivalent of hydrogen [53]. However, there
is still some residual intensity from NH3 groups indicating the reaction is not complete. With the
50AB-50PEO sample the NH3 peak is totally removed by the heating and the resulting peak is
representative of PAB although narrower than for the heated AB. The dehydrogenation in the
50AB-50PEO material is faster than in pristine AB at 85°C so over the same period of time (six
hours) we expect to see less NH3 in the 50AB-50PEO material than in pristine AB.
Ex situ heating
Samples of the electrospun 50AB-50PEO fibre mat composite where encased in baking foil,
placed in small test-tubes, and then held in an oil bath at 85°C for distinct periods of time. The
heating time ranged from 10 to 180 minutes. On removal, the samples were quenched by plunging
the test-tube into iced water, a technique that rapidly reduced the temperature to halt the
dehydrogenation.
-320 -340 -360 -380
NH NH2
heated
heated
pristine AB
ppm
50wt% AB fibres
NH3
97
The fibres, clearly visible in the as-spun 50AB-50PEO material, agglomerate swiftly at 85°C as
after 10 minutes no fibres are in evidence. This is expected as PEO in the 50AB-50PEO fibres
melts at 76.5°C. The melt has flowed together to fill the gaps between the fibres. The samples are
very thin as evidenced by the holes through the material and when handling them they were
observed to be very brittle. The initial three images of the fibres that have been heated for 10 20
and 30 minutes respectively, exhibit two distinct regions a rough surface and below it a smooth
undulating space. It is likely that during the heating the melted material flows into the vacant
space that exits between the fibres, additionally it will seek to reduce its overall surface area.
When running the experiment, firstly aluminium foil was used to package the fibres as it provides
an inert surface against the sample; unfortunately it adhered strongly to the melted PEO and the
sample was impossible to remove. Baking paper was substituted which improved the situation;
however there is still some stickiness between the paper and the PEO which is confirmed as being
the cause of the rough surface. Between 1 and 2 hours the hydrogen release causes the once
smooth craters to become pockmarked with tiny holes or likely gas vents.
With higher magnification the crater surface appears coarser rather than full of holes, this could
be due to the AB having changed to PAB and the craters themselves are evidence of large gas
bubbles.
98
Figure 52: The changing surface of the 50AB-50PEO electrospun fibres at 85°C.
50AB-50PEO
20 min
60 min
30 min
90 min
10 min
99
Figure 53: Higher resolution image of pockmarked
Heating XRD
In the XRD pattern (Figure 54) both AB and PEO are present as evidenced by the AB tetragonal
crystal structure (110) and (101) peaks and the PEO triplet at 22.5° and the quartet at 27° peaks
(see Figure 34). After 10 minutes under 85°C heating, a slight peak shift is observed, probably an
artefact of the melted and recrystallized process. After 20 minutes under 85°C heating the sample
can return to its crystal form. At 30 minutes the peak features have all but disappeared to be
replaced by an amorphous hump at 22°. This amorphous peak indicates a large proportion of the
AB in the 50AB-50PEO material has reacted to PAB [190]. Of course this hump must also
represent all the PEO in the sample as well, as there are no distinctive peaks it can be assumed
that the AB and PEO are well mixed.
Figure 54: X-ray diffraction pattern of the 50AB-50PEO electrospun fibres heated ex situ over 3 hours.
The peaks associated with AB and PEO disappear and are replaced with an amorphous hump. A hump at
22° has previously been associated to PAB.
10 15 20 25 30 35
Inte
nsi
ty,
a.u
.
2
50AB-50PEO
10 min
20 min
30 min
60 min
90 min
180 min
100
Discussion and conclusions
Structure of the ammonia borane-polyethylene oxide composites
Electrospinning the AB-PEO solutions as described, produced piles of integral fibres. Some
melding of the fibres occurred at cross points indicating that the fibres were not completely dry
when they reached the collection plate. The low AB content fibres (25AB-75PEO) were smooth,
similar to the pure PEO fibres. Increasing the AB content (50AB-50PEO and 75AB-25PEO)
caused the fibres to become rough and porous.
The diffraction patterns of the AB-PEO electrospun fibres showed that both AB and PEO are
present in the electrospun fibres. The pristine AB has a tetragonal structure which was maintained
in the 75AB-25PEO and 50AB-50PEO samples but not in the 25AB-75PEO sample. The pristine
PEO is monoclinic. In the electrospun materials, new peaks, as well as the traditional monoclinic
ones, are observed. This suggests some, if not all, of the PEO is distorted by the presence of AB.
The diffraction data imply that a new AB-PEO phase has formed.
The vibrational energies of the bonds in the pristine AB and electrospun PEO fibres were observed
via FTIR and corresponded to that reported in the literature. When the AB and PEO were
combined in the AB-PEO electrospun fibres, new reflections occurred. These new bands suggest
a shift form AB-to-AB dihydrogen bonding to AB-to-PEO hydrogen bonding as the PEO content
in the fibres increase. Additionally structural changes in the PEO backbone were also observed.
Further, the nitrogen NMR performed suggests that the new hydrogen bonding in the 50AB-
50PEO sample is between the HN on AB and the O in the PEO. The boron NMR signal of the
50AB-50PEO suggests that this new hydrogen bonding stabilises a portion of the AB in the AB*
form.
Foaming of ammonia borane and of the ammonia borane-polyethylene oxide
composites
The tendency of AB to produce foam while releasing hydrogen, limits its applicability. The
hydrogen release of the pure AB in Figure 39 is preceded by an endothermic melting curve,
meaning that the gas must pass through a highly viscous liquid, characterised by strong
intermolecular interactions, to escape. These are ideal circumstances for bubbles. AB can release
hydrogen slowly in the solid phase and more quickly in the liquid phase.
In the AB-PEO composites, the foam production is successively restricted, as the PEO content is
increased. It is therefore safe to conclude that PEO is responsible for supressing the foam.
101
In Figure 39 it is noted that the melting of AB is curbed by the presence of PEO and that the
hydrogen release occurs at lower temperatures in the solid phase. If the AB is not melting, there
is no liquid to contain the released gases, and so less foaming will occur.
High density pellets expand more than low density ones. Consequently, free volume within the
compressed material lessens foam development. This suggests that the interconnected space
provides escape routes for the hydrogen and that more free volume minimises gas containment
and the build-up of pressure in the pellet.
PEO is used as an industrial antifoaming agent as it contains both oily and polar domains. The
polar domains/areas mix well with other polar materials and connect via dipole-dipole
interactions. In contrast, the oily areas only exhibit weak intermolecular interactions with like
species. When bubbles form, the oily domains constitute the weakest point in the bubble wall,
helping it to break sooner (at lower internal pressures) and so control the build-up of foams. As
the foam tests were conducted at 120°C, it is safe to assume that the AB is liquid. Also, the NMR
and IR experiments show polar hydrogen bonding interactions that intimately connect the AB to
the PEO. This suggests that in the composites, the PEO is an integral part of the bubble wall which
causes the bubble to burst at lower internal pressures and therefore subdue the foaming.
The PEO used in the pellets has a chain length of 2 million monomer units. The integral fibres
formed in the electrospinning show the PEO chains to be entangled. Once the fibre mat is
compressed into a pellet, the AB is effectively tied together by an interconnected polymer
framework. This framework would hold the AB together and work against the expansion caused
by the foaming.
These arguments are not mutually exclusive and may explain different contributions to the
antifoaming influence of the PEO on AB.
Hydrogen release
The pristine AB releases hydrogen under ramp heating in two steps corresponding to the 1st and
2nd hydrogen release steps observed in the literature. With successive addition of PEO to the AB,
the onset temperatures for these steps are lowered. This suggests the PEO has a positive effect on
AB by reducing the energy barrier to dehydrogenation.
The AB-PEO fibres were also analysed under step heats to a variety of set temperatures. At each
temperature, earlier release of hydrogen, with a reduced induction period, was directly related to
102
the PEO content in the fibres. This suggests that PEO improves both the kinetics and
thermodynamics of the hydrogen release from AB.
The dehydrogenation of the samples was accompanied by mass loss, the majority of which occurs
during the release of the 2nd H2 Equiv. Even though the mass loss profile (in all cases)
corresponded to the hydrogen trace in the RGA, the total mass loss exceeded that accounted for
by hydrogen by at least three times. The total mass loss from the pristine AB and the AB-PEO
fibres is similar. This indicates the PEO affects the mass loss rate but is not responsible for the
excess of mass loss quantity. Other gases such as water, borazine etc. or sublimed solids, could
be responsible for the unexpected change. This implies a filter would be required to purify the gas
stream before feeding the hydrogen into a fuel cell.
The changing DSC trace profile suggests that increasing the amount of PEO in the material
enables the AB to release significant quantities of hydrogen below its melting temperature.
In the AB-PEO composites greater levels of borazine were observed in the hydrogen gas stream
than compared with the pristine AB. Also, the borazine was observed at lower temperatures in
the gas stream from the fibres (corresponding to the release of the 2nd H2 equiv.). This strongly
implies that PEO causes AB to release more borazine than in its pristine state.
Many studies have previously been conducted on AB dehydrogenation in the presence of liquid
ether type compounds, diglyme, tetrglyme, THF etc. [78]–[80], [191]. These studies showed
accelerated hydrogen releases but also large amounts of boron containing products. As well as
the chemical effect of the ether group, the liquid state of AB was also assumed to be improving
the hydrogen release. This work demonstrates that the hydrogens release can be accelerated whilst
suppressing the melting of AB. This indicates that the catalytic effect of the group is more
important than the physical state of AB at the time of hydrogen release.
Reaction mechanism in the gas stream
There are increased levels of borazine in the gas stream observed from the AB-PEO fibres as
compared to AB. The borazine and diborane are observed at earlier times in the AB-PEO fibres
(in accordance with the release of 2nd H2 equiv.), and also the overall levels are higher. This
suggests that PEO is encouraging the AB to release hydrogen via the cyclic reaction route.
The in situ boron NMR shows that the PEO massively increases the speed of the reaction. In both
the pristine AB and the 50AB-50PEO material, the AB first becomes mobile (AB*) before
forming DADB. This means that PEO lowers the activation energy to the formation of DADB.
AB is stabilised by its hydrogen bonding network [176]. The IR data shows that in the 50AB-
103
50PEO sample, the bending modes N-H and B-H are changed in the presence of PEO. This
indicates that the AB is forming hydrogen bonds to oxygen in PEO. An overall reduction in AB-
AB hydrogen bonding would push the AB-DADB equilibrium towards DADB. An alternative
way to think about the system is that DADB is a more polar molecule then AB and that DADB
would prefer the polar environment supplied by the PEO more than AB [66].
In the 50AB-50PEO, the BH4 reflection is broader than in AB. Also, throughout the reaction, in
the 50AB-50PEO, the BH4 reflection shifts to higher energy than in AB. In pristine AB, BH4
reflection shift is suggested as due to the reaction of DADB (highly shielded) to PAB (reduced
shielding). (A shift to higher energy in a material indicates a reduction in the shielding effect of
the electron cloud around the boron, suggestive of longer B-H bonds.) There are three
possibilities: one the BH4 is in the vicinity of cyclic moieties in preference to PAB, two, the
oxygen is interacting with the boron, three, the tetrahedral arrangement of the hydrogens has been
disturbed, possibly by forming a bridge hydrogen with the boron end of an adjacent AB molecule.
One is unlikely as a good proportion of the AB does react to PAB as seen in the XRD pattern of
the heated 50AB50PEO material. Two is also unlikely as BH4 is negatively charged overall, so
the oxygen would donate its electron pair elsewhere, perhaps to the BH2. The third possibility is
the most likely.
One of the greatest differences between the 50AB-50PEO sample compared to pristine AB are
the reflections due to B-O interactions. The B-N bond in AB is dative as the N donates into the
vacant B orbitals; it is highly probable that the B would also accept electrons from other donating
species close by, namely the O.
The FTIR of 50AB-50PEO sample shows that hydrogen is lost from both the nitrogen and the
boron end of the AB molecule. Also, it confirms the presence of B to O bonds throughout the
dehydrogenation.
Further work
Structure
This thesis has concluded that PEO and AB in the composites are well mixed, that they interact
via hydrogen bonding and that a new AB-PEO phase is formed. The structure of this new phase
is however unknown. Further work to determine the crystal structure could include refinement
of the XRD pattern guided by molecular dynamics simulations.
104
Foaming
The reduction in foaming of the AB in the PEO composites during dehydrogenation has been
linked to reduced melting of the AB, weakening of the bubble wall and structural integrity of the
material. Further work could include viscosity measurements on molten pristine AB and the
composites and a detailed study of AB-PEO composites with different chain length PEO
polymers.
Impurities
The high level of boron containing impurities in the gas stream from the AB-PEO composite,
does not make PEO an attractive additive for AB. Further work should include the study of
polymers with different catalytic functional groups.
Reaction Mechanism
The AB in the composites follows both the polymeric and cyclic reaction route. The cyclic route
is encouraged by the PEO but no hydrogen is lost from the polymer. The barrier to DADB
formation is lowered and boron to oxygen bonds are formed. This is still very poorly understood
more research will be needed. Simulation work would be extremely helpful as would selective
deuteration of the B and N ends of the AB molecule before performing, IR, NMR and RGA in
situ studies.
105
5 Polystyrene and ammonia borane
Electrospinning polystyrene and ammonia borane
Polystyrene (PS) has been extensively electrospun in the past in a variety of solvents including
dimethylformamide (DMF), tetrahydrofuran (THF), t-butylacetate, chlorobenzene, chloroform,
dichloroethane, 1,4-dioxane, MEK, and toluene to name but a few [107], [131], [137], [192]–
[194]. A blend of equal parts by mass of DMF and THF was selected. AB decomposes slowly
in THF [50] so diluting with DMF is prudent. DMF has a low vapour pressure, diluting with THF
raises the vapour pressure sufficiently to produce dry fibres. The DMF-THF blend is ideal for
single phase electrospinning as it is a co-solvent for both PS and AB [103], [195].
PS of two weight average molecular weights, 350,000g mol-1 (350k) and 1,000,000g mol-1 (1M),
were tested. Various concentrations and spinning conditions were explored to isolate the ideal
parameters for the system (Table 17). A suitable regime for the stable spinning of the pure PS
solution was identified before AB was added. This initially limits the number of variable
parameters and is an efficient way to determine the ideal spinning conditions for a new solution.
Once stable spinning of a pure PS solution had been achieved AB was added to give a the required
dry mass AB:PS ratios (Table 18).
Table 17: The spinning conditions tested for the PS in a 1:1 ratio of THF:DMF to determine the ideal
system parameters before adding AB.
Polymer
Mw
g/mol
Solvent Conc.
wt%
Flow
rate
ml/hr
p.d.
keV Observations
PS
350K
1:1
DMF:THF
10 2
8-30
Terrible, multiple jets formed.
15
1-5
Better at 2.5-5ml/hr with 5keV. But still
excess jets observed.
20 Stable spinning achieved at 2.5-5ml/hr with
15keV. This system is ideal.
25 Stable spinning achieved at 2.5-5ml/hr with
15keV.
PS 1M
1:1
DMF:THF
1
0.5-10
8-30
Stable at any flow rate. 15keV ideal voltage.
3
Fabulous at 15keV, flow rate must be
<2.5ml/hr to stop multiple jets forming.
Makes shiny fibre mat.
5
Fabulous at 15keV, flow rate must be
<2.5ml/hr to stop multiple jets forming. Makes
shiny fibre mat.
106
Table 18: The spinning conditions tested for the AB-PS single phase solution to determine the ideal
system parameters to produce stable spinning conditions and dry fibres.
Polymer
Mw
g/mol
Solvent Conc.
wt%
AB:PS
ratio
Flow
rate
ml/hr
p.d.
keV Observations
PS 1M
1:1
DMF:THF
3
50:50
0.5-10
8-30
0.5-1ml/hr multiple jets! But flow
rate between 2.5-10ml/hr much
improves the system! Very fluffy
fibre mat produced. 5ml/hr with
12keV gives stupendously stable
spinning!
60:40 5ml/hr under 15keV, produced dry
fibres.
70:30 5ml/hr under 15keV, produced dry
fibres.
80:90 5ml/hr under 15keV, produced dry
fibres.
90:10 5ml/hr under 15keV, produced dry
fibres.
During the electrospinning of the AB-PS fibres, bubbles were formed in the syringe
(approximately 2cm3 volume). This indicates that AB was beginning to decompose in the THF
probably because the solution was not completely anhydrous. This distorted the flow rate and the
reduced slightly the final amount of AB in the fibres. Minimising both the amount of AB in the
solution and the time AB spends in the solvated state limited bubble formation. This was achieved
with a low AB content and a high flow rate. To reach a high AB to PS ratio, the PS content in the
solution must also be low. With a longer chain PS polymer (1M) far lower concentrations (i.e.
1wt%) are sufficient to afford the viscosity and chain entanglement necessary for fibre production.
Finally, 3wt% PS solution spun at 5ml/hr flow rate was chosen as a compromise between fast
sample production and averting bubble formation when AB was added.
Five distinct AB-PS fibre materials were produced containing nominal AB contents of 50, 40, 30,
20 and 10wt%, these are termed 50AB:50PS, 40AB:60PS, 30AB:70PS, 20AB:80PS and
10AB:90PS respectively. The 50AB:50PS fibres where the first created and tested. No particular
lowering of the dehydrogenation of the AB in this sample was observed. In chapter 4, the
temperature of the dehydrogenation of AB was lower as the AB content in the AB-PEO fibres
was lowered. Therefore, further AB-PS fibres were electrospun, with low AB contents to
determine if a similar effect would be observed here.
On the collection plate, the pure PS fibres formed a flat, dry and shiny mat below the nozzle. The
fibres made from the PS and AB containing solutions did not lie down flat but instead amassed
into a fluffy, cotton wool like, pile. The cotton wool effect is likely due to static charge in the
fibres. As this effect was also noticed when electrospinning the PEO-AB solutions it is likely that
the AB is responsible for the charge retention.
107
The vertical build-up of PS-AB fibres resulted in a reduction of the whipping distance over time.
This meant, the initial fibres produced had more inflight drying and stretching time than those
produced later. AB decomposes in THF it is prudent to assume the first fibres will have a different
AB content then those produced an hour later from the same solution. This implies that there is
variability across the fibre sample.
Figure 55: SEM image of PS and 50AB:50PS electrospun fibres.
The electrospun fibres have a ribbon like shape and highly crinkled surface (see Figure 55). This
is somewhat unusual as PS fibres are usually rounder and smoother with surface pores caused by
humidity in the environment [144], [192], [196], [197]. However, short PS chain lengths, 100-
500K, and high solution concentrations, 25-35wt%, are favoured in these studies. Wrinkles can
form when an outer skin, caused by fast surface evaporation, collapses slowly inward as the core
dries [134], [197]. The low polymer concentration, 3wt%, afforded by the extensive polymer
chain length (1M) meant a high proportion of the jet cross section was solvent. Therefore, as the
large solvent volume dries out, the skin shrinks a long way inwards resulting in many creases.
The PS fibres are 2-3 µm in diameter while the 50AB:50PS are larger, 5-10µm. For the PS sample,
the flow rate was kept below 2.5ml/hr while for the 50AB:50PS solution, 5ml/hr was used. The
potential field was 15keV in both cases. An increase in conductivity usually results in a decrease
in fibre diameter [107]. The AB containing solution is more conductive then the pure PS solution,
but it has been electrospun at a higher flow rate. This indicates that for the PS-AB solution the
high conductivity has been counteracted by the higher flow rate to give wider fibres.
Foam tests on the ammonia borane-polystyrene fibres
From each of the five AB-PS fibre samples, three pellets were made for foam testing. (Only two
pellets where made from the 30AB-70PS fibres as there was insufficient material for three.) The
AB-PS pellets were heated in a standard foam test (oil bath at 120°C). The AB in the PS fibres
foamed less than pristine AB. Also, as the PS content was raised a corresponding sequential drop
PS 50AB-50PS
108
in the foaming was observed (see Figure 56). In addition to the foam, some solids have detached
from the pellets and coated the internal top half of the test-tubes. Again, as the PS content is
raised, the amount of solid residue drops.
The high AB content sample (50AB-50PS) expands and foams as a whole in all directions,
reminiscent of the 75AB-25PEO pellet. The median AB content samples (40AB:60PS and
30AB:70PS) expand somewhat before the AB bubbled out of the surface of the pellet. The low
AB content pellets (20AB:80PS and 10AB:90PS) elongated along the vertical axis, the direction
in which the compression force was applied. This suggests that the electrospun PS has formed a
flexible network that the foaming AB can swell. When the maximum distortion of the PS is
reached the AB foam continues growing (if sufficient AB is in the sample) and bubbles out of the
pellet surface.
Figure 56: Foam testing AB-PS pellets made from the electrospun fibres. Foaming and residue on the test
tube wall is inversely related to the PS content. Images of the final pellets once removed from the test
tube have also been inserted.
This implies that PS is not actively responsible for the suppression of foaming in the samples.
The high PS containing samples foam less, as the actual AB content and density in the samples,
is low.
Ramp heating of ammonia borane-polystyrene fibres
The AB-PS samples were tested under a standard ramp heating run in a combined
thermogravimetric analyser (TGA) and differential scanning calorimeter (DSC) with attached
mass spectrometer (mass spec). The TGA and DSC data has been normalised to total sample
Bef
ore
A
fter
50AB-50PS 40AB-60PS 30AB-70PS 20AB-80PS 10AB-90PS
109
mass and the mass spec data has been normalised to the nominal mass of AB in the sample and
the argon level (see experimental chapter for more details).
Figure 57: TGA trace of the mass loss of the AB-PS fibres when subject to a 2°C/min temperature ramp
between room temperature to 200°C.
Figure 58: DSC traces of the AB-PS electrospun fibres under ramp heating rate of 2°C/min.
The total mass loss increases with the nominal AB content in the fibres in Figure 57. The mass
loss profile is similar for the pristine AB and the AB-PS fibres. However, some of the AB-PS
fibres, particularly the 50AB-50PS sample show mass loss to begin at a slightly lower temperature
then pure AB.
80 100 120 140 160 180
60
80
100
Mas
s lo
ss, %
Temperature, oC
Pure AB
50AB-50PS
40AB-60PS
30AB-70PS
20AB-80PS
10AB-90PS
80 100 120 140 160 180
0.0
0.5
1.0
1.5
Hea
t fl
ow
, W
/g
Hea
t fl
ow
, W
/g
Temperature, oC
80 100 120 140 160 180-2
-1
0
1
2
3
Pure AB
50AB-50PS
40AB-60PS
30AB-70PS
20AB-80PS
10AB-90PS
110
Figure 59: The hydrogen and impurity trances for the AB-PS fibres. When combined with PS, AB
releases hydrogen at lower temperatures than when in its pristine state. The PS does not appear to have a
significant effect on the impurity levels.
The DSC curve profiles of the AB-PS fibres in Figure 58 are similar to that of pristine AB;
endothermic dip indicating melting followed by two exothermic peaks for the release of the 1st
and 2nd hydrogen equivalents (H2 equiv.). Yet, the graph shows some differences. Firstly, the
peak of the first exothermic dip occurs at a slightly lower temperature in the AB-PS fibres than
in the pristine AB. And secondly, the endothermic melt dip is shallower as compared to the
exothermic peak in the AB-PS fibres than in the pristine AB. This suggests that hydrogen release
starts earlier in the AB-PS fibres than in pristine AB, before the AB has melted.
Let us take a quick look at the impurities that accompany the hydrogen release form the AB-PS
fibres. In Figure 59 the peaks, for the release of the 1st H2 equiv. from the AB-PS samples, reached
their apex around 104°C, a good 8°C before the pristine AB which peaks at 120°C. The ammonia
and borazine graphs also show that the AB-PS samples react at lower temperatures than pristine
AB. The actual impurity levels in the hydrogen gas stream cannot be accurately determined from
this plot. However, the plot implies that while the AB in the fibres decomposes at lower
80 100 120 140 160 180
0.0
0.2
0.4
0.6
Am
monia
sig
anl,
m/z
=17,
Dib
ora
ne
sigan
l, m
/z=
2,
BorazineDiborane
AmmoniaHydrogen Pure AB
AB1_1PS
AB4_6PS
AB3_7PS
AB2_8PS
AB1_9PS
80 100 120 140 160 180
0.000
0.005
0.010
0.015
0.020
Temperature, 0C
Temperature, 0C
norm
alis
ed t
o m
ass
of
AB
and a
rgon f
low
, a.
u.
norm
alis
ed t
o m
ass
of
AB
and a
rgon f
low
, a.
u.
80 100 120 140 160 1800.000
0.002
0.004
0.006
0.008
80 100 120 140 160 1800.000
0.001
0.002
Bora
zine
sigan
l, m
/z=
81,
Hydro
gen
sig
anl,
m/z
=2,
111
temperatures then the pristine AB the impurity levels do not vary significantly across the AB-PS
samples.
Some difference between the dehydrogenation of pure AB and AB-PS fibre samples as a whole
is observed. However, no meaningful trend connecting the AB-PS samples is observed. This
suggests that the PS itself does not have an active effect on the dehydrogenation of AB. Rather
something else in the composites is responsible.
Conclusions and discussion
Here it is shown that PS and AB can be electrospun into homogeneous fibres from a single
solution. The fibres produced are flat and wrinkly likely due to the low polymer content.
The foam tests show that successively adding PS to AB via electrospinning successively supresses
the foaming. However, the AB in the fibres behaves in a fashion similar to bulk AB. This
indicates that the reduced foaming is in fact the result if the reduce amount of AB in the materials.
The dehydrogenation of the AB in the AB-PS fibres is initiated at slightly lower temperatures
than the pristine AB. However, no trends are observed when the PS content in the AB-PS fibres
is varied. This suggests that the PS itself is not responsible for lowering the AB reaction
temperature.
Perhaps, some of the AB in the fibres is nanostructured. As AB is polar molecule and PS a
nonpolar polymer it is likely that they would repel each other in solution and heterogeneous bulk
regions would occur in the solid fibres. Yet, it is possible that some AB has become isolated in
PS nano-cavities [100] and it is these nano-sized AB particles that react first and initiate hydrogen
release in the bulk sections of the AB at lower temperatures. This theory seems somewhat
farfetched and it is possible the AB was activated in a different way.
All the solutions had the same PS content, 3wt%, and were electrospun with the same kit under
the same conditions (5ml/hr at 15keV). The amount of AB was the only changing factor. In all
the cases, during the electrospinning gas was evolved inside the syringe and this was identified
as the AB decomposing in the THF [79]. This means, the AB in the fibres is likely contaminated
with some small amounts of partially reacted AB such as DADB or PAB. It has been shown
previously that adding DADB to AB significantly reduces the induction time to hydrogen release
[56]. As all the electrospun AB remained in solution for a similar period of time it is possible that
a similar percentage of AB was converted to DADB in each case and distributed within the bulk
AB.
112
The foaming and dehydrogenation behaviour of the electrospun AB-PS fibres shows that
nanostructured AB was not achieved. It is suggested that the lowering of the induction
temperature of the AB in the fibres is due to the processing parameters. AB in this form is not
suitable as hydrogen store for a fuel cell as a 50% weight concession has been made and the
hydrogen release properties have not been significantly improved.
Further work
Combining AB with PS via electrospinning does not, according to the data presented here,
ameliorate the hydrogen release properties of AB. Further work on this system is therefore not
recommended.
113
6 Ammonia borane and clay
AB was combined with several distinctive clays; montmorillonite (MONT), laponite (LAP),
imogolite (IMOG) and halloysite (HAL). These composite materials were prepared via freeze
drying and contain equal amounts by dry mass of AB and clay (1:1 AB:clay). Freeze drying limits
the carrier solvent choice to one – water to which 5wt% clay was added. The IMOG and LAP
dispersed well and formed gels but the HAL and MONT did not dissolve and sedimented on the
base of the jar. The gel from the LAP and IMOG clays necessitated pre-dissolving the AB, so for
consistency, all the solutions were made by mixing the clay into an AB solution.
The HAL used in the experiments was dry powder, as HAL is not amenable to rehydration this
could have be the cause of its inability to dissolve. [116]. In the case of MONT, while the layers
swell, the AB water mix was not suitable to totally exfoliate the layers. This is probably due to
the high interlayer charge that binds the layers together [198].
Scanning electron microscopy
The microstructure of the four AB-clay composites was studied by SEM and the images are
displayed in Figure 60. The AB-LAP and AB-IMOG composites exhibit dramatic structural
change as compared to the pure clay while in the case of AB-MONT and AB-HAL composites,
little contrast is observed.
The LAP clay is synthetic and of high purity which is reflected in the regular, square edged
particles observed under the SEM. The freeze dried LAP-AB material is porous and flaky with a
cracked, pitted surface. When AB is added to IMOG, the material produced is composed of thin,
transparent flakes that taper to points. The HAL tube bundles are well recognised in both the pure
and the composite material. The only noticeable difference is the increase in grain size when AB
is added. This suggests that the AB is binding the HAL tubes together. In the case of MONT, no
differences between the clay and composite are observed.
These images indicate that the IMOG and LAP have become well mixed through the freeze drying
process with the AB. The HAL and MONT have not blended well with the AB, most probably as
a result of the unsuccessful dissolution.
114
Figure 60: SEM images of the four AB-clay composites. In the case of IMOG and LAP, the significant
morphology change between the as received clay and the AB-clay composite, suggests intimate mixing of
the clay and AB has been achieved.
HAL-AB
MONT-AB
IMOG-AB
LAP-AB
IMOG
LAP
HAL
MONT
115
Foam tests of AB-clay composites
Elimination of the foaming and expansion associated with the H2 release from the AB is key to
its commercial prospects and so the foam test is the first test a new material must pass. Pellets
where made in the usual way from the 1:1 AB-clay freeze dried materials and the foaming
response on heating was tested with the standard foam test method previously described in the
experimental chapter. In Figure 61 a reduction in foaming is observed for all the clay-AB
composites. Two contrasting effects were noticed. Foam and expansion, with some large bubbles,
are observed in the HAL and MONT composites, while a more explosive change, with a greater
solid loss but fewer bubbles, is noted in the LAP and IMOG materials.
Figure 61: AB:clay 1:1 ratio pellets post foam testing.
The LAP and IMOG pellets could be removed post heating while the HAL and MONT samples
adhered to the internal wall of the test-tube. The LAP pellet was extremely fragile, crumbling
instantly while the IMOG version was swollen but maintained its integrity, see Figure 62. This
suggests that the tubular nature of the IMOG can provide a stronger framework than the disk
shapes that are LAP.
LAP IMOG HAL MONT
116
Figure 62: The LAP and IMOG AB-clay pellets post foam testing. The AB-LAP pellet instantly crumbles
while the AB-IMOG remains whole.
In Figure 63 the change in pellet density, volume and mass which occurred during the foam test
are displayed. The change in pellet mass (the final mass includes residue on the internal wall of
the test tube) caused by the heating is high (around 25%) and fairly constant across the samples.
It can be inferred that the mass loss is independent of clay type. The H2 loss can account for a
maximum of 13.1% mass loss [36] and the excess is probably due to the sublimation of solids and
the discharge of heavier gaseous elements. The change in volume on the other hand varies hugely
from clay to clay. The AB-MONT pellets displayed the greatest swelling (600%) and the AB-
IMOG pellets outperformed the other composites with the smallest volume expansion (150%).
Figure 63: The change in mass, density and volume of the AB-clay pellets as a percentage of the initial
dimensions post foam testing.
LAP IMOG
117
In Figure 38 in Chapter 4 the AB-PEO composites demonstrated a direct relationship between the
initial density of the pellet and the expansion on heating. The initial densities of the AB-clay
pellets are in general higher than their PEO counterparts, AB-clay>1g/cc>AB-PEO, and this is
attributed to the higher density of clay, ~2g/cc [109] when compared to polymers ~1g/cc. For this
reason it is to be expected that the clay composites would foam more than the AB polymer
materials but this is not the case, see Figure 64. Also, in Figure 64, with the data available, no
correlation between expansion and initial density that could explain the various volume changes,
is evident. The points for the AB-IMOG sample hint at a positive correlation but the data are
limited and no clear assertion can be made.
Figure 64: Volume expansion against initial density of the AB-clay composites. The AB-IMOG is the
only sample that suggests a positive correlation.
The complete destruction of the pellets of the AB-MONT and AB-HAL composites suggests that
AB is in the bulk state. Some abatement of foaming, as compared to pure AB is however observed.
Perhaps the presence of the clay creates more escape pathways for the gas so less pressure can
build in the pellet. Or, the clay separates the AB clusters and on melting, the AB cannot form
large interconnected bubbles. Another possibility is that the solid state and high mass of the clay
makes it harder to be displaced.
In the AB-IMOG and AB-LAP pellets expansion is limited and the initial pellet shape is
maintained. However, in place of foaming, matter is ejected from the surface. This indicates that
while the clay and AB are strongly bound together, the integrity of the pellet is limited due to the
small particle size of the clay. This suggests that as hydrogen is liberated, pressure builds up in
the pellet. Finally the hydrogen forces its way out carrying lose matter with in the gas stream.
118
Ammonia borane-imogolite: varying the proportion of AB
The AB-IMOG mix provides the most desirable properties in terms of foaming, and perhaps
IMOG would be a suitable additive to ameliorate the hydrogen release properties of AB. To
determine quantitatively the effect of IMOG on AB three new composites were made with
AB:IMOG ratios 1:3, 2:2 and 3:1. The AB content in solution was varied while the IMOG ratio
in the gel was kept constant. The samples were 1AB-3IMOG, 2AB-2IMOG and 3AB-1IMOG for
the AB-IMOG 1:3, 2:2 and 3:1 ratios respectively.
Figure 65: AB-IMOG pellets after foam testing.
In Figure 65 it is clear that the AB-IMOG pellets all foamed less than pure AB. The 3AB-1IMOG
sample only demonstrated a minor decrease in foaming as compared to the pure AB. The 1AB-
3IMOG and 2AB-2IMOG pellets however maintained their original shape only swelling
minimally. The heated 2AB-2IMOG pellets here were more brittle than those from the first batch
[XREF fig 2], perhaps due to the initial compression force, but the solid deposits on the internal
test-tube surface are similar in both cases. Combining IMOG with AB has a successively positive,
limiting influence on the foaming of AB.
SEM images of the three, uncompressed AB-IMOG composites were taken to determine if
changes in the microstructure could be responsible for the change in foaming, see Figure 66. The
flakes that give the 1AB-3IMOG a corallesque appearance disappear as the proportion of AB
increases. In the 3AB-1IMOG sample, the amount of porous surface is reduced and this suggests
a reduction of free volume in the composite.
1AB-3IMOG 3AB-1IMOG 2AB-2IMOG Pure AB
119
Figure 66: SEM image of the 1AB-3IMOG, 2AB-2IMOG and 3AB-1IMOG freeze dried composites. As
the proportion of AB increases the visible free volume decreases.
X-ray diffraction of AB-IMOG composites
X-ray diffraction (XRD) patterns were obtained for the composites. All data were collected on
the Rigaku using an automatic sample changer apart from the AB which was collected on the
Phillips X’pert. Firstly, by comparing the position of the reflections from with the sample holder,
the 75AB-25IMOG sample was observed to be slightly displaced to higher angles (Figure 63).
The samples are coarse powder and each was glued to a different but similar sample holder. This
suggests that the displacement is likely the result of varied sample height in the beam. The pristine
AB peaks are also displaced but to higher angle. In this case the shift is probably due to the data
being collected on the Phillips X’pert rather than the Rigaku.
Figure 67: The two main AB tetragonal peaks for pristine AB and IMOG. The relative displacement is
likely due to the difference of the height of the sample as shown in the displacement of the sample holder
peak.
1AB-3IMOG 2AB-2IMOG 3AB-1IMOG
64 651
10
100
1000
Inte
nsi
ty,
a.u
.
2 theta
22 23 24 25 26 27
100
1000
10000
Inte
nsi
ty,
a.u
.
2 theta
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
IMOG - as-received
Instrument
(110)
(101)
120
AB is observed in all composites and is structurally unchanged by the presence of IMOG. The
peak intensity changes attest to the changing proportions of AB present in Figure 68. The slight
shifts observed in the AB peak positions are due to variation in the sample height level under the
beam.
Figure 68: The AB peaks in the composites are observed at the same 2 theta position.
The diffraction pattern for the initial IMOG sample has an exceptionally broad peak at 5°
(between 5 and 20Å) and sharper reflections at 19.8, 28.3, 35, 53.5, 61.0 and 72.5° (corresponding
to 2.25, 1.35, 1.61, 0.95, 0.85 and 0.8Å) see Figure 69. The broad span, around 5°, encompasses
the literature values for the (001) repeat distance along the IMOG tube (8.25Å) and the centre to
centre distance of two adjacent tubes (~20Å) [114], [199]–[201]. The peak being broad suggests
an ill-defined crystalline arrangement for the IMOG tubes, probably amorphous [202]. The peaks
55 60 65 70 75 80
0
200
400
600
800
1000
1200
1400
Inte
nsi
ty, a.
u.
2 theta
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
IMOG -as recieved
sample holder
(130)(013)
30 35 40 45 50 55
0
200
400
600
800
1000
1200
1400
Inte
nsi
ty,
a.u
.
2 theta
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
IMOG - as-received
sample holder
(100)(101)
(220) (002)
(121) (112)
30 35 40 45 50 55
0
200
400
600
800
1000
1200
1400
Inte
nsi
ty,
a.u
.
2 theta
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
IMOG - as-received
Instrument
(100)(101)
(220) (002)
(121) (112)
121
between 19 and 72.5° could not be identified, but have previously been associated with IMOG
[200].
Figure 69: XRD pattern of the AB IMOG freeze dried composites.
As AB is successively added to the IMOG the broad peak around 5° (5 to 20Å) becomes sharper.
A better defined peak position is observed in the 25AB-75IMOG sample at 8.5° and in the 50AB-
50PEO sample at 7°. In the 75AB-25PEO the peak has almost vanished. The reduction in intensity
around 8° suggests that the IMOG tubes become isolated from each other due to the presence of
AB as shown in Figure 70. The AB and IMOG must be well dispersed within each other.
Figure 70: The freeze drying process used to prepare the AB-IMOG composite leaves the IMOG tubes
dispersed within the AB.
0 20 40 60 8010
100
1000
10000
Inte
nsi
ty,
a.u
.
2 theta
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
IMOG - as-received
Instrument
AB dissolved
in H2O IMOG added,
gel formed
H2O extracted,
porous solid
remains
Stirred Frozen
122
Thermogravimetric analysis and differential scanning calorimetry of the
ammonia borane-imogolite composites
The dehydrogenation characteristics of the three AB-IMOG composites was studied via standard
ramp heating runs in the combined TGA and DSC with attached mass spec. The IMOG seems to
have a positive effect on the hydrogen release characteristic of AB. In Figure 71 the peaks
corresponding to the release of the first and second hydrogen equivalents (1st and 2nd H2 equiv.),
both move down in temperature as the proportion of IMOG increases. Concurrently, the hydrogen
release ‘pre-peak’ at 85°C intensifies. This indicates that hydrogen is released at lower
temperatures in the composites than the pristine AB.
Figure 71: Hydrogen release profile of the AB-IMOG composites compared to pure AB under a standard
ramp heating. As the IMOG content increases the pre-peak increase and the release of the 1st and 2nd
hydrogen equivalent (H2 equiv.)
In Figure 72 the TGA mass change curve profile switches from a two-step to a three-step process,
as the fraction of IMOG in the sample increases. The gradient inflection points correspond to the
DSC peak maxima, showing they are directly linked to the exothermic events in the sample. The
DSC peak maxima in Figure 72 also correspond to the hydrogen release peak maxima in Figure
71. The mass loss can thus be attributed to the decomposition of the AB in the samples.
60 80 100 120 140 160 1800.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
2nd
H2 equiv.
1st H
2 equiv.
Hy
dro
gen
sig
nal
, m
/z=
2,
no
rmal
ised
to
mas
s o
f A
B a
nd
arg
on
flo
w,
a.u
.
Temperature, oC
AB - as received
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
pre-peak
123
Figure 72: TGA and DCS traces for the 1AB-3IMOG, 2AB-2IMOG and 3AB-1IMOG freeze dried
composites under standard ramp heating.
In Figure 72, the change in the 1st and 2nd H2 equiv. release peaks is modest when compared to
the large intensity change in the pre-peak. Looking at the DSC curve in Figure 72 we can see that
this hydrogen is being released before the AB melts (melting point of AB is between 95°C and
105°C). This indicates that the IMOG enables the AB to lose hydrogen while in the solid phase.
The shape of the curves suggests that two types of AB exist in the composites, one in the bulk
that behaves like standard AB and a second that is closely involved with the IMOG, let’s call it
AB#. AB# is probably in direct contact with the IMOG.
Hypothetically as the temperature is raised, AB# will release hydrogen before the bulk AB
accounting for the pre-peak. The reaction front will propagate outwards [54] and the DADB and
PAB evolved at this stage, would go on and react with the adjoining bulk AB before the natural
AB melting temperature is reached. This would bring down the temperature of the release of 1st
and 2nd H2 equiv. peaks. Finally, the increasing sample temperature passes 100°C and causes any
remaining bulk AB to melt and react in the normal way. If the 1st H2 equiv. peak moves down in
60 80 100 120 140 160 180-2
-1
0
1
2
3
60 80 100 120 140 160 180
0.5
0.6
0.7
0.8
0.9
1.0
AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
DS
C n
orm
alis
ed
to A
B c
onte
nt,
mW
/mg
Mas
s ch
ange
norm
alis
ed
to A
B c
onte
nt,
a.u
.
Temperature, oC
Temperature, oC
124
temperature the associated exotherm will overshadow the endotherms associated with the bulk
AB melt. Thus the melting endotherm in the DSC decreases as the percentage of IMOG is
increased. This model accounts for the pre-peak, the earlier appearance of the 1st and 2nd H2 equiv.
release peaks and the decrease in melting endotherm as a function of increasing IMOG.
There is one further detail, the 2nd H2 equiv. release peak seems to be actually representing two
different overlapping peaks: one at 140°C and the other at 154°C. This is most evident in the
2AB-2IMOG sample. The dehydrogenation of AB can be described as a chain reaction [53], [54].
Assuming, the composite contains bulk AB and AB# and that the AB# reacts at lower
temperatures, two peaks for the release of the 2nd H2 equiv. are too be expected.
Hydrogen gas stream content
Besides the hydrogen release, the impurity levels must also be considered. For this, cumulative
integrals of the mass spec traces detected under the heating ramp from the pure AB and the AB
IMOG composites have been plotted (Figure 73). It is difficult to compare the pure AB to the AB-
IMOG composites for the reasons outlined in Chapter 3. The AB traces have been included
because, while the intensity cannot be trusted, the temperatures at which gradient changes occur,
is valid.
On examination of the three clay composites, some trends can be observed. Firstly, the proportion
of ammonia in the hydrogen stream increases as more IMOG is added. Conversely, the levels of
diborane and borazine drop with greater amounts of clay. Water levels though remain fairly
constant. All samples have been dried in the same freeze dryer over the same period of time and
it is expected that the water residue would be similar. Even so the 1AB-3IMOG sample releases
water between 60°C and 90°C whereas the other two composites only perceptibly release water
after 100°C. Perhaps this in linked to the greater clay content or this is a feature of the measuring
equipment.
125
Figure 73: Cumulative integrals of the mass spec traces of hydrogen and impurities detected under the
heating ramp. The data are normalised to the hydrogen (m/z = 2) level at 190ºC and errors in the data are
shown as the lighter background. As the IMOG content is raised more ammonia is released but borazine
and diborane are decreased.
20 40 60 80 100 120 140 160 1800.0
0.2
0.4
0.6
0.8
1.0
1.2
m/z
=2
, n
orm
alis
ed
Temperature, oC
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
Hydrogen
40 60 80 100 120 140 160 180
0.00
0.01
0.02
0.03
0.04Ammonia
m/z
=18,
norm
alis
ed t
o h
ydro
gen
sig
nal
m/z
=81,
norm
alis
ed t
o h
ydro
gen
sig
nal
m/z
=27,
norm
alis
ed t
o h
ydro
gen
sig
nal
m/z
=17,
norm
alis
ed t
o h
ydro
gen
sig
nal
40 60 80 100 120 140 160 1800.000
0.005
0.010
0.015
Temperature, CTemperature, C
Temperature, C
Temperature, C
Diborane
40 60 80 100 120 140 160 1800.000
0.001
0.002
0.003
0.004
Water
Borazine
40 60 80 100 120 140 160 180
0.00
0.05
0.10
0.15
0.20
Pure AB
3AB-1IMOG
2AB-2IMOG
1AB-3IMOG
126
Discussion and conclusions
Structure
The x-ray diffraction patterns suggest that the IMOG tubes are well dispersed while the AB has
managed to dry into its preferred tetragonal crystal structure, even in the 1AB-3IMOG sample.
This indicates that while the AB and IMOG are not mixed and have formed well dispersed regions
of homogeneous AB and IMOG. Likely these bulk clusters of AB have boundary regions of AB#
that are connected to the IMOG.
IMOG is a tubular clay with an internal diameter of around 10Å which water is able to penetrate
[114], [200], [203], [204]. When the IMOG is immersed in the AB–water solution it is possible
that as well as solvating the surface, the water entered into the tubes, dragging AB along.
Theoretically, one could assume that some of the AB is within the tubes. The diffraction pattern,
however, attests to a tetragonal AB crystal and there are no unexplained peaks associated with
AB nor any broad humps to signify amorphous areas. This suggests that the tubes are AB free
and that AB# is likely connected to the external hydroxyls.
Foaming
When IMOG and LAP where intimately combined with AB, the foaming of AB was dramatically
reduced. When HAL and MONT where mixed with AB, the foaming was not as much suppressed.
This indicates that the clay morphology, layered (MONT and LAP) or tubular (HAL and IMOG)
did not significantly affect the foaming of the composites. The highest foam reduction was
achieved with the clays (LAP and IMOG) that formed gels with the AB-water mix. In a gel the
clay particles will be interspersed with the AB molecules as shown in Figure 70. On drying this
will ensure AB is integrated with the clay. This suggests that the foaming of AB is controlled by
intimate mixing with clay.
AB could be facilely blended with IMOG in a variety of ratios, via freeze drying, provided the
AB was dissolved before the IMOG. The fraction of IMOG in the composite was observed to be
inversely related to the amount of foaming that occurs. The microstructure of the AB-IMOG
materials as viewed via SEM suggests that the IMOG is responsible for adding porosity to the
solids. As pores provide escape routes for the gas, with more pores, less hydrogen could be
trapped in the material and less foaming would occur. Alternatively, the increasing proportion of
IMOG in the sample could serve simply to increase the separation between AB clusters. If the
AB clusters are dispersed, the chance of forming large stable bubbles reduces.
127
In the DSC the AB showed reduced melting as the IMOG content was raised. The reduced
foaming could be due to AB releasing more hydrogen in the solid state. Less liquid AB reduces
the likelihood of foam occurring.
Dehydrogenation
The IMOG content in the composites has a direct effect on the quantity of ammonia in the gas
stream. The majority of the ammonia is released between 80°C and 95°C, coinciding with the
‘prepeak’ of the hydrogen release. To release ammonia from AB the B-N bond must be severed.
The IMOG surface, Al-Oδ--Hδ+ could attract the boron side of AB, Bδ+-Hδ-. Electron transfer to
the boron could weaken the B-N bond till it breaks and liberates ammonia. This would leave space
for hydrogen molecule to form, half from the IMOG and half from the boron resulting in a B-O
bond such that Al-Oδ- Bδ+-H2δ+, please see Figure 74. H2 release from AB has previously been
shown to be initiated boron interaction with an active oxygen i.e. carbon cryogel [84], [87] [89].
Zhao et al showed how a B-O bond can lead to the simultaneous release of H2 and NH3 [60].
Figure 74: Hypothetical reaction scheme for the initiation (prepeak) of hydrogen release from AB by the
hydroxyl groups on the IMOG surface.
The step of the release of the 1st H2 equiv. occurs between 95°C and 120°C. Hydrogen is the main
gas released from the composites over this temperature range and it is accompanied by a small
amount of ammonia. A possible mechanism for this step, in Figure 74, shows how in place of
DADB the dehydrogenation of AB can be catalysed by BH2NH3+.
The release of the 2nd H2 equiv. occurs between 130°C and 160°C. Here, the hydrogen release is
accompanied by borazine and diborane. However, as the IMOG content in the composite
increases the fraction of borazine and diborane in the gas stream reduces. This indicates that the
AB follows the polymeric dehydrogenation route (ABPABPIB) as opposed to the cyclic one
(ABCTBborazine) [46]. PAB produced in the 1st step has reacted to PIB in the 2nd step. As
the IMOG fraction increases more hydroxyl groups are available to AB (making AB#) which then
follows the polymeric AB dehydrogenation route.
The borazine and diborane traces are limited but not eliminated by the clay. This suggests that
areas of bulk AB that release hydrogen in the traditional way are present in the composites. This
Clay
128
assertion is supported by the diffraction pattern which shows AB tetragonal crystal reflections
even in the 25AB-75IMOG composite.
IMOG improves the kinetics of hydrogen release from AB, lowers impurities (bar ammonia) and
controls foaming. The hydroxyl surface is actively involved in the reaction as the hydrogen is
replaced by a stronger bond to the boron. A 50% weight penalty (50AB-50PEO) is necessary to
significantly control the foaming. To significantly speed the hydrogen release (25AB-75IMOG)
a 75% weight penalty was required. While the IMOG does provide some hydrogen from the
hydroxyls, the release of 1 hydrogen atom from a hydroxyl is accompanied by the loss of 1
ammonia molecule. Each ammonia equates to the loss of 2 useable hydrogen atoms, this in total
is therefore a net loss of 1 hydrogen atom per hydroxyl group that interacts with AB.
The ammonia could be cleaned from the gas stream by adding an ammonia absorber such MgCl2
[108] and could be used to feed a polymer membrane fuel cell [205]. However, this would
necessitate a further weight penalty perhaps unnecessary when the material could be used to feed
a combined hydrogen and ammonia solid oxide fuel cell.
Further work
Structure
FTIR and molecular dynamics simulations, for example of an AB molecule on an IMOG surface,
could be employed to determine the nature of the suggested interaction between the AB and
IMOG as suggested. The IMOG and AB crystallite sizes could be determined Scherrer analysis
of the XRD peaks and energy dispersive X-ray spectroscopy.
Foaming
The data suggests that the foaming is reduced in the composites as the decomposition of AB is
following a different reaction route that initiates before the AB melting temperature is reached.
In situ microscope studies could be useful to study how and when the bubbles, if any, form. Also
high resolution SEM combined with TEM could be employed to see if the study the AB particle
sizes.
Reaction
More conclusive evidence is required to determine if the AB is actually reacting with the IMOG.
This could be obtained by deuterating the AB and then observing the masses of the released
hydrogen moieties with the mass spec. Selective deuteration of one side of the AB molecule
129
would give more in depth information. In situ IR, Raman, inelastic neutron scattering and NMR
studies of the heated AB-IMOG composites would also be useful.
Alternative materials
It was shown here that in all likelihood the AB has not crystallised inside the IMOG tube. A
synthetic IMOG with wider tube diameter could perhaps encourage the AB into the channel and
increase the proportion of active surface available to the AB.
The AB-LAP composite was dropped from further experimentation due to the structural
weaknesses, expansion and ejected solids, exhibited in the foam test. However, studying the
effect of the LAP on the hydrogen release behaviour of AB, under perhaps a more moderate
heating regime, is still valid. Comparisons between a layered clay (LAP) and tubular clay (IMOG)
could be made.
If MONT and HAL clays could be fully solvated, an intimate blend of clay and AB could be
created. Solvents that could be tested are salt water and THF [198]. As freeze drying is most
suitable for water, electro spraying could be used to process the solutions or gels made with
alternative solvents into a dry state.
This leads on the studies on alternative clays. Many and varied clay materials could be combined
with AB and tested as to their suitability as hydrogen store. Metals, transition and alkali earth,
have been shown to have a catalytic effect on AB dehydrogenation and these also could be
included in the composites [91]–[94]
130
7 Neutron spectroscopy studies of calcium-ammonia solutions
confined by graphene sheets.
Synthesis
Madagascan flake graphite was intercalated with calcium (Ca) to form a Ca graphite intercalation
compound (GIC) using the single zone metal vapour technique [33], [149]. The raw materials,
graphite and excess calcium were placed into a steel container and held under vacuum at 465°C
for 10 days; during this time the calcium vaporised and intercalated into the galleries.
1.16g of the Ca GIC was packed in a cylindrical aluminium sample can, 50mm high by 20mm
diameter and mounted in the neutron beam. A gas manifold containing a 0.5l ammonia vapor
reservoir was fitted and a high vacuum pumping system attached to the sample can. The sample
was evacuated to a pressure of 10-7 mbar and subsequently exposed to an NH3 atmosphere. The
NH3 was loaded slowly by raising the pressure from 0.5 to 5.2 bar at 300K over approximately 14
hours. Before readings commenced the pressure was reduced to 3bar pressure to ensure only the
residual material was measured this material is termed Ca-NH3 GIC. Afterwards, the reversibility
of the NH3 intercalation was examined by holding the pressure at 10-5 mbar at 300K for 1hour.
The high temperature and low pressure conditions were used to encourage the loosely bound NH3
to deintercalate.
The quasi elastic neutron scattering (QENS) spectrum was recorded at 50K intervals moving from
300K to base at 2K. The analysis focused on the energy range -0.3 to 0.8 meV, beyond 0.8 meV
the scattering becomes asymptotic to the baseline. Scattering was observed over the Q range from
0.42 to 1.85 Å-1 and analysed by segregating the detectors into 17 groups. Neutron diffraction
patterns were measured between d = 2.7 and 7.2Å. The IRIS wavelength bands are narrow, so to
achieve the entire spectra displayed the Bragg reflections were collated from the pyrolytic
graphite 002 and the mica graphite analysers. The data were analysed with MODES v3 a program
designed specifically to process data from the IRIS beam line [169].
Diffraction
In the diffraction pattern of the as-intercalated Ca-NH3 GIC in Figure 75 (purple) two peaks were
observed, one at 3.355Å assigned to the graphite (002) [24], [26] and one at 4.518Å assigned to
layer spacing of the Ca GIC. The d-spacing Ca GIC layer of stage 1 CaC6 is quoted in literature
as 4.6Å [164], [206], [207]. The d-spacing measured here for the Ca GIC layer is 4.518Å and a
stage 1 arrangement is likely considering the gold colour of the sample [26]. The peak at 4.518 Å
131
is broader than the graphite (002) peak, this suggests that the Ca intercalated areas may be smaller
than the graphite ones.
Figure 75: The neutron diffraction pattern pre ammoniation (purple line) shows both Ca GIC and graphite.
After ammoniation (long red trace) a new multistaged ternary phase Ca-NH3 GIC was created. The final
Ca-NH3 GIC sample includes stages 2, 3 and 4 labelled GIC2, GIC3 and GIC4 respectively. During the
NH3 intercalation unidentified high stages (marked *) form before the low stages.
Table 19: The neutron diffraction pattern shows the Ca-NH3 GIC to have multiple stacking arrangements.
Stage 2, 3 and 4 are present but no stage 1 is observed. This is due to the low NH3 pressure of 3bar which
limits the amount of NH3 in the galleries.
00l reflection Stage 1(Å) Stage 2 (Å) Stage 3 (Å) Stage 4 (Å)
001 6.4 [206] 9.52 ‡ 12.87‡ 16.25‡
002 4.76§
003 3.17§ 4.29§
004 3.22§ 4.06§
005 3.246§
§ Observed distances and ‡ calculated distances from observed peaks.
The difference in peak intensity between the graphite (003) peak and the Ca GIC suggests more
graphite layers than Ca intercalate layers. When the NH3 is added the Ca become solvated and
require more pristine graphite to expand into. At low NH3 pressure, 0.5 bar, a broad peak (marked
*) developed around 3.3Å. The peak became more pronounced at 1.4bar and when the pressure
was raised to 4.0bar a second peak appeared at 3.28Å. These peaks (marked*) could not be further
identified without data at higher d-space. The ammoniation is accompanied by a reduced intensity
3.1 3.2 3.3 3.40
100
200
300
400
500
600
700
800
5 6 70
100
200
300
400
500
600
700
800
3.5 4.0 4.5 5.0
Counts
, a.u
.
*
C
ounts
, a.u
.
C-Ca empty
C-Ca-0.5bar NH3-10min
C-Ca-0.5bar NH3-50min
C-Ca-1.4bar NH3-100min
C-Ca-1.4bar NH3-130min
C-Ca-5.2bar NH3-160min
C-Ca-5.2bar NH3-250min
C-Ca-5.2bar NH3-1230min
C-Ca-3.0bar NH3-1580min
C-Ca, NH3-pumped off
(002) Graphite
(003) GIC2
(004) GIC3
(005) GIC4 (004) GIC4
(003) GIC3
Ca in layer
(002) GIC2
*
d-spacing, Å 3.1 3.2 3.3 3.40
100
200
300
400
500
600
700
800
5 6 70
100
200
300
400
500
600
700
800
3.5 4.0 4.5 5.0
Cou
nts,
a.u
.
*
Cou
nts,
a.u
.
C-Ca empty
C-Ca-0.5bar NH3-10min
C-Ca-0.5bar NH3-50min
C-Ca-1.4bar NH3-100min
C-Ca-1.4bar NH3-130min
C-Ca-5.2bar NH3-160min
C-Ca-5.2bar NH3-250min
C-Ca-5.2bar NH3-1230min
C-Ca-3.0bar NH3-1580min
C-Ca, NH3-pumped off
(002) Graphite
(003) GIC2
(004) GIC3
(005) GIC4 (004) GIC4
(003) GIC3
Ca in layer
(002) GIC2
*
d-spacing, Å
132
of the graphite (003) reflection suggesting the solvated Ca is infiltrating the pristine graphite even
at low NH3 pressures.
After stabilising the sample under 3bar NH3, additional reflections appeared at the expense of the
Ca GIC and graphite peaks which diminished. With reference to Srinivas et al and Hönhe et al,
the reflections were indexed to the 00l reflections of a ternary Ca-NH3 GIC with various stage
arrangements; stages 2, 3 and 4 are evident. Measured reflections and the associated calculated
values for the d-spacing distance of the various stages are shown in Table 19. No stage 1 ternary
GIC is apparent. Since the staging of the system is controlled by the applied ammonia pressure
[207], [208] and stage 1 can be achieved if the pressure is raised to around 7 bar [209] we assume
a low ammoniation pressure, 3bar, is insufficient to produce stage 1. Unidentified stages, defects
or diffuse scattering from within the layers, may be responsible for the broad feature at 3.5Å as
the majority (80%) of the coherent scattering is due to the graphite Table 10.
Figure 76: The layers of the Ca-NH3 GIC contract when the temperature decreases from 300K to 2K.
The interlayer spacing of the final Ca-NH3 GIC structure at 3bar NH3 is sensitive to temperature.
In Figure 76 all the layers contract when the temperature is reduced from 300K to 2K. This is due
to the reduction in thermal motions as temperature decreases. Also, the ammoniated layers put
pressure on the adjoining graphite layers forcing them together. In figure X the graphite (001)
layer at 3.355Å decreases to 3.353Å when NH3 it is added. To accept NH3 into the galleries the
graphene layers must expand which will result in strain on the graphite [26]. This data suggests
that to minimise the strain the graphene layers move closer together.
0 50 100 150 200 250 300
3.10
3.15
3.20
3.25
3.30
3.35
(002) graphite - no NH3
(002) graphite - with NH3
(005) GIC4
(004) GIC3
(003) GIC2
d-s
pac
ing, Å
Temperature, K
133
Diffusional dynamics
The motion of the ammonia molecules was studied at ~50K intervals between 300K and 2K. The
data were refined by removing a background (including the sample can) and fitting with an elastic
line curve (Gaussian) and one or two quasi scattering curves (Lorentzian) as appropriate.
Figure 77: In graph A the total neutron scattering spectrum of the Ca-NH3 GIC is plotted as a function of
Q for a range of representative temperatures. The fit to the data are good as demonstrated by a small
residual. The spectra broaden and the peak height drops as the temperature increases. In graph B the
elastic incoherent structure factor, EISF, ratio of the elastic scattering over the total scattering for all Q,
against temperature is depicted. With increasing temperature the shift from elastic to inelastic scattering is
due to the protons becoming more active and indicates that greater motion is available to the ammonia.
Considering the sharp drops in the trace of the right-hand graph, we can preliminarily define the motional
modes occurring.
In Figure 77 the total scattering spectrum over a range of representative temperatures is depicted.
As the temperature is raised the peak are under the peak and the width increases. This is the result
of an increasing fraction of the beam being scattered by the sample and indicates that a greater
range of motion possible for the ammonia at higher temperatures. Previous studies have identified
a range of possible modes of motion of ammonia in similar situations beginning with rotation
-0.05 0.00 0.05
0
2
4
6
8
10
12
14
16
0 50 100 150 200 250 3000.6
0.7
0.8
0.9
1.0
300K
200K
S(Q
,)
Energy transfer, meV
100K
A Total neutron
scatering
Residual
Combined Lorenztian
and Gausian fit
B
EIS
F
Rotation
Temperature, K
Static
Diffusion
-0.05 0.00 0.05
0
2
4
6
8
10
12
14
16
0 50 100 150 200 250 3000.6
0.7
0.8
0.9
1.0
300K
200K
S(Q
,)
Energy transfer, meV
100K
A Total neutron
scatering
Residual
Combined Lorenztian
and Gausian fit
B
EIS
F
Rotation
Temperature, K
Static
Diffusion
-0.05 0.00 0.05
0
2
4
6
8
10
12
14
16
0 50 100 150 200 250 3000.6
0.7
0.8
0.9
1.0
300K
200K
S(Q
,)
Energy transfer, meV
100K
A Total neutron
scatering
Residual
Combined Lorenztian
and Gausian fit
B
EIS
F
Rotation
Temperature, K
Static
Diffusion
134
about a fixed position at low temperatures, hopping from one metal site to another at medium
temperatures and diffusing freely at high temperatures [121]–[123], [210]. A plot of the elastic
line intensity (elastic incoherent structure factor, EISF) against temperature in Figure 77 suggests
distinctive modes of motion take place at different temperatures. The large change in the
amplitude between 250K and 200K and the second shift between 100K and 50K demonstrates
that the type of movement of the ammonia is experiencing changes at these points.
Figure 78: The FWHM of the Lorentzian fits to the data at 300 and 250K, previously divided into 17
groups to depict the changes as a function of Q2. The Singwi-Sjölander jump model provides the best fit
to the data describing an exponential distribution of jump lengths and frequency around the mean values.
At 300K this corresponds to a jump distance of 3.2±0.1Å with a frequency of 19ps and at 250K, this
corresponds to a jump distance of 2.88±0.2Å with a frequency of 28 ps.
At both 300K and 250K a Gaussian and single Lorentzian was found suitable to meet the
statistical requirements of the fit. In Figure 78the FWHM is plotted as a function of Q2 and in both
the 300K and 250K cases exemplifies the Singwi-Sjolander jump diffusion model. At 300K
ammonia is typified as jumping 3.2±0.1Å every 19ps. The motion is retarded at 250K with a
shorter jump of 2.9±0.2Å that occurs less often, every 28ps. As the sample cools, less energy is
available for motion accounting for the variance in the jump profile at 300K and 250K. The
diffusion coefficients, at 300K and 250K, reflect this and are calculated as 8.91×10-5 cm2s-1and
4.02×10-5cm2s-1 respectively. Considering a molecular graphics snap-shot of CaC12(NH3)2 in
Figure 79 the jump distances calculated with the Singwi-Sjolander model are seen to be realistic.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.50
50
100
150
200
Q2, Å
-2
FW
HM
,
eV
300K
200K
Singwi - Sjolander fit
135
Figure 79: Molecular graphics snap-shot of a single layer of composition CaC24(NH3)2 showing the jump
lengths expected (Unpublished work from Neal Skipper).
A Delta and two Lorentzians where required to fit the spectra acquired at the temperatures of
200K, 150K and 100K. The fitted parameters are Q independent (Figure 80) and indicate a
rotational mode. The determined values for the rotational diffusion constant and rotational
dynamic correlation times are displayed in Table 20. A point has been discounted in the fit of the
data at 200K at Q2 = 0.92Å-2 as the error bar spanned the entire energy range. A large residual
accompanied the fittings at 200Kand we suggest some diffusion is also present as observed by
Neumann et al in KC24(NH3)4.3 [211]. The scattering is however dominated by a rotation signal.
Figure 80: In graph A the FWHM are shown with respect to Q2 at 200K, 150K and 100K. The linear
nature of the fit suggests rotational modes with higher energy at higher temperatures. In graph B an
Arrhenius plot of the FWHM obtained yields the activation energy off the rotation as 696 Jmol-1.
1.42Å
2.30Å
2.8-3.3 Å
2.7Å
0.004 0.006 0.008 0.010
5.4
5.6
5.8
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.50
200
400
600
ln(F
WH
M),
eV
FWHM
Linear fit
1/T, K-1
200K
150K
100K
Linear fit
F
WH
M,
eV
Q2, Å
-2
A B
136
Table 20: The FWHM, rotational dynamic correlation times, 𝜏𝑟𝑜𝑡, and rotational diffusion constant, Drot,
calculated at temperatures of 100K, 150K and 200K.
Temp. (K) 200 150 100
FWHM (μeV) 322 ±9 272 ±3 212 ±3
𝜏𝑟𝑜𝑡 (ps) 0.68 0.81 1.05
Drot(μeV) 161 136 106
The second Lorentzian peaks for the temperatures 200K, 150K and 100K are within the resolution
of the instrument and therefore cannot be analysed. A spectrometer with a higher resolution is
required to discern if another higher order rotational mode is present or if this is merely a feature
of the background. At 0 and 50K no spreading into the QENS was detected. While some
movement may still be occurring again it is not observable without an instrument of higher
resolution.
Figure 81: The QENS spectrum before ammoniation, during the experiment and after the ammonia was
pumped off, the change in intensity demonstrates the partial reversibility of ammonia intercalation.
Once the readings had been taken the pressure in the sample chamber was reduced down to ~10-
5 bar while the temperature was gently raised from 2K to 300K over approximately 15 hours in
an attempt to deintercalate the ammonia. In Figure 81the intensity of the QENS spectrum is raised
due to the ammonia intercalation and drops when the ammonia pressure is removed. However,
the final peak does not return to the same level as the initial Ca GIC. This indicates residual
ammonia is present within the structure and shows the intercalation process is only partially
reversible. Possibly when the pressure is removed molecules that are near the sample edges will
escape, while others remain trapped in the structure perhaps in intercalant islands or defect sites
[26].
-0.05 0.00 0.05
0.00
0.05
0.10
0.15
0.20
0.25
S(Q
,)
Energy transfer, meV
Initial CaC6
3bar NH3
Final pumped off
137
Conclusions and discussion
A binary Ca GIC (CaC6) was generated via the vapour transport method. A ternary NH3-Ca GIC
was then made in situ by an over pressure of 5.2 bar NH3 atmosphere. The ternary GIC
composition was stabilised at 3bar NH3 and found by diffraction to be multi-staged. The process
of ammoniation initially created high stage ternary phases followed by lower stages as the level
of ammonia was raised. While the exact composition of the ternary GIC is unknown the interlayer
NH3-Ca arrangement will be constant through the sample and each NH3-Ca complex will have a
maximum of four surrounding NH3 [34], [118], [119]. The interlayer graphite space is 3.353Å, it
increases with the intercalation of Ca to 4.518Å and then further on addition of NH3 to 6.16Å.
All of these distances are temperature dependent however, as when the sample is taken to 2K a
marked decrease in the spacing is observed. This indicates that the structure contracts as the
temperature falls.
Neutron scattering was used to elucidate two modes of movement and one still state. Hopping
diffusion following the Singwi-Sjolander model occurred at 300K and 250K and as the sample
cooled the hopping movement disappeared to be replaced with a rotational mode at temperatures
200K, 150K, and 100K.
The diffusion coefficients at 300K and 250K have been calculated as 8.91×10-5 cm2s-1 and
4.02×10-5 cm2s-1 respectively for CaC12(NH3)2, it is expected that less energy is available for
motion at lower temperatures. In the K analogue, KC24(NH3)4.3, at 300K, the translational
diffusion of the same order of magnitude 10-5 cm2s-1 was observed, however this was
accompanied by a rotational mode of 1.17meV that was not observed in CaC12(NH3)2 [211].
At 200K, 150K and 100K the rotation of NH3, successively decreasing in energy is observed.
These reorientation times (0.68 and 1.05ps) are within the same order of magnitude as those
observed in the octahedral Ca-NH3 complex (1.0ps at 120K) [120] suggesting that the interaction
energy between the Ca and ammonia is unaffected by the containment in the layers.
However, compared to the K-NH3 GIC studies by Neumann et al [211], the energy of rotation of
the NH3 in the Ca-NH3 GIC is much higher (270μeV compared to 130μeV at 150K). I suggest
this is the direct result of the dipole increase between the metal and the ammonia due to the change
from the univalent K to the divalent Ca. At 50 and 2K no movement was observed in the Ca GIC,
this is contrary to both literature sources which detect broadening indicative of rotation at 40K in
Ca(NH3)6 [120] and at 78K in KC24(NH3)4.3 [211].
The ammoniated state is only partially reversible at conditions of 300K and ~10-5 bar. The final
trace in the diffraction graph, figure three, shows the sample after deintercalating has been
138
attempted. Two broad peaks, one at 3.2Å and the other at 3.26Å are present, similar to what was
observed by Srinivas et al. [206]. The graphite (001) peak has also shifted down slightly to
3.347Å. The disappearance of intensity at 3.17Å suggests that the stage 2 GIC, the lowest in this
sample, disappears first. The peak broadness suggests an irregular array of filled and empty layers.
A plausible model is that NH3 is indiscriminately pulled out of the edges of the structure leaving
a haphazard arrangement of stages behind.
Future work
While this study has characterized the high temperature behaviour of the ammonia in the Ca
GIC, at 50K and below the data are poor. In studies of Ca-NH3 solution [120], [212] and a Cs-
NH3 GIC [210] tunnelling rotations of the hydrogen atoms are observed below 50K. A high
resolution neutron scattering spectroscopy study with longer counting times or hydrogen nuclear
magnetic resonance spectroscopy are both suitable probes. Molecular dynamics simulations could
provide insight into the exact rotation and hopping mechanisms involved.
139
8 Final conclusions
Structure and foaming of the ammonia borane composites
The tendency of AB to produce foam while releasing hydrogen limits its applicability. The
hydrogen release of the pristine AB is preceded by an endothermic melting curve. This means
that the gas must pass through a highly viscous liquid, characterised by strong intermolecular
interactions, to escape. These are ideal circumstances for bubbles. AB can release hydrogen
slowly in the solid phase and more quickly in the liquid phase.
All the composites synthesised foamed less than pristine AB. Additionally increasing the additive
content in the sample successively impedes the foaming. The PEO, PS and clays are very distinct
both in structure and chemical behaviour. This suggests that simply disrupting the long range
connectivity of the AB to itself reduces the foaming.
Let us consider the 50AB-50additive materials first. The foaming and expansion varied depending
on the additive. The PS-AB pellets showed the greatest expansion, for the AB-clay pellets the
volume increase was more than 100% while the AB-PEO pellets expanded by less than 100%.
The XRD pattern of the 50AB-50PEO suggests the presence of an AB-PEO mixed phase. In the
50AB-50IMOG material, the AB is observed by the XRD in its tetragonal crystalline state. Let
us consider the material from a theoretical perspective. Assuming the AB indulges in hydrogen
bonding with either electron donors (ether oxygen in PEO) or electron acceptors (IMOG
hydroxyl) hydrogen bonds can be formed with the IMOG and PEO but not to the same extent
with the PS.
The DSC traces for the 50AB-50additive (PEO, PS and IMOG) composites suggested that the
foam reduction could be linked to a decrease in the melting of AB. When less liquid is available
less foam can form.
In the AB-PEO pellets, expansion and foaming showed a positive correlation with initial pellet
density. This suggests that the interconnected space provides escape routes for the hydrogen and
that more free volume minimises gas containment and the build-up of pressure in the pellet.
Hydrogen release from the ammonia borane composites
Pristine AB releases hydrogen in two steps, corresponding to the 1st and 2nd hydrogen release, that
take place around 110°C and 150°C. This is ideally high and limits the applicability of AB as a
solid state hydrogen storage material in a vehicle.
140
With successive addition of PEO and IMOG the AB, the onset temperatures for these steps are
lowered. Adding PS has no appreciable effect on the dehydrogenation temperature of AB. This
implies that the PEO and IMOG have a positive effect on AB by reducing the energy barrier to
dehydrogenation. In the AB-PS fibres the AB decomposed exactly like bulk AB, in the AB-PEO
fibres the AB decomposed like bulk AB (two steps) but these occurred at lower temperatures and
in the AB-IMOG composite a portion of the AB reacted like bulk and a portion reacted via an
alternative route as suggested by the ‘pre-peak’.
The impurity release was also quite illuminating. The impurities in the gas stream from the AB
in the AB-PS fibres resembled that of pristine AB. In the AB-PEO fibres the AB released
increased amounts of boron containing impurities. And in the AB-IMOG composite has increased
levels of ammonia, but reduced boron containing products. The impurity levels in the hydrogen
stream the AB-PEO and the AB-IMOG were directly related to the fraction of additive in the
material. This suggests that the chemical nature of the additive environment was responsible. This
implies that ether groups foster the production of boron containing by-products, a statement that
is in line with the sources literature quoted in the materials introduction and that hydroxyl groups
are responsible for increased ammonia levels, not commonly observed in the literature.
Solid state hydrogen storage
So now the final question, are any of these composites suitable to be used as a hydrogen storage
system for a vehicle? Well the AB-PS system is clearly not suitable in any way. The 25AB-
75additive composites (PEO and IMOG) are able to release hydrogen quickly at lower
temperatures than pristine AB. However, the weight penalty combined with the impurity release
does not fit these materials for purpose. Further work is required.
Calcium graphite intercalates
A multiphase ternary calcium ammonia GIC was made by intercalating a mix of stage 1 CaC6 and
graphite with pressurised ammonia vapour. The intercalating material moved through a vast
number of phases indicating radical reordering and diffusion of the interlayer species. The
calcium ammonia GIC was found to be highly sensitive to temperature. Thermal expansion of the
layers was observed and the diffusion of the interlayer ammonia was affected. Hopping diffusion
following the Singwi-Sjolander model occurred at 300K and 250K and as the sample cooled the
hopping movement disappeared to be replaced with a rotational mode at temperatures 200K,
150K, and 100K. The mechanism by which the GIC restages is unknown.
141
9 Bibliography
[1] M. Hirscher, Handbook of Hydrogen Storage: New Materials for Future Energy Storage.
Weinheim: WILEY-VCH Verlag GmbH & Co. KGaA, 2010.
[2] S. Satyapal, J. Petrovic, C. Read, G. Thomas, and G. Ordaz, “The U.S. Department of
Energy’s National Hydrogen Storage Project: Progress towards meeting hydrogen-
powered vehicle requirements,” Catal. Today, vol. 120, no. 3–4, pp. 246–256, Feb. 2007.
[3] R. H. Jones and J. G. Thomas, Eds., Materails for the hydorgen economy. Boca Raton:
CRC Press Taylor & Francis Group, 2008.
[4] G. Walker, Ed., Solid state hydorgen storage Materails and chemistry. Cambridge
England: Woodhead Publishing Limited and Maney Publishing Limited, 2008.
[5] J. O. M. Bockris, “The hydrogen economy: Its history,” Int. J. Hydrogen Energy, vol. 38,
no. 6, pp. 2579–2588, Feb. 2013.
[6] A. Zuttel, A. Borgschulte, and L. Schlapbach, Hydrogen as a Future Energy Carrier.
Weinheim: Wiley-VCH Verlag GmbH & Co, 2008.
[7] IEA, “ENERGY,” 2013.
[8] M. Mohseni, Bahram Ramezanzadeh, H. Yari, M. Moazzami, and M. M. Gudarzi, New
Advances in Vehicular Technology and Automotive Engineering. In-Tech, 2012.
[9] L. Carrette, K. a. Friedrich, and U. Stimming, “Fuel Cells - Fundamentals and
Applications,” Fuel Cells, vol. 1, no. 1, pp. 5–39, May 2001.
[10] IEA, “Hydrogen Production & Distribution,” 2007.
[11] DOE, “Targets for Onboard Hydrogen Storage Systems for Light-Duty Vehicles,” pp. 1–
22, 2009.
[12] “Driving the Future Today A strategy for ultra low emission vehicles in the UK,” London,
UK, UK, 2013.
[13] W. I. F. David, “Effective hydrogen storage: a strategic chemistry challenge,” Faraday
Discuss., vol. 151, p. 399, 2011.
[14] A. Züttel, A. Remhof, A. Borgschulte, and O. Friedrichs, “Hydrogen: the future energy
carrier.,” Philos. Trans. A. Math. Phys. Eng. Sci., vol. 368, no. 1923, pp. 3329–42, Jul.
2010.
[15] D. J. J. Sloan, Clathrate hydrates of natural gases. New York: Marcel Dekker, Inc., 1990.
[16] R. Chahine and T. K. Bose, “Low pressure adsorption storage of hydrogen,” Int. J.
Hydrogen Energy, vol. 19, no. 2, pp. 161–164, 1994.
[17] P. Ciambelli, D. Sannino, M. Sarno, a. Fonseca, and J. B. Nagy, “Selective formation of
carbon nanotubes over Co-modified beta zeolite by CCVD,” Carbon N. Y., vol. 43, no. 3,
pp. 631–640, Jan. 2005.
142
[18] G. Yushin, R. Dash, J. Jagiello, J. E. Fischer, and Y. Gogotsi, “Carbide-Derived Carbons:
Effect of Pore Size on Hydrogen Uptake and Heat of Adsorption,” Adv. Funct. Mater.,
vol. 16, no. 17, pp. 2288–2293, Nov. 2006.
[19] A. Dailly, J. J. Vajo, and C. C. Ahn, “Saturation of hydrogen sorption in Zn
benzenedicarboxylate and Zn naphthalenedicarboxylate.,” J. Phys. Chem. B, vol. 110, no.
3, pp. 1099–101, Jan. 2006.
[20] B. Panella and M. Hirscher, “Hydrogen Physisorption in Metal-Organic Porous Crystals,”
Adv. Mater., vol. 17, no. 5, pp. 538–541, Mar. 2005.
[21] H. Chae, D. Siberio-Pérez, J. Kim, and Y. Go, “A route to high surface area, porosity and
inclusion of large molecules in crystals,” Nature, vol. 427, no. February, pp. 523–527,
2004.
[22] N. B. McKeown and P. M. Budd, “Polymers of intrinsic microporosity (PIMs): organic
materials for membrane separations, heterogeneous catalysis and hydrogen storage.,”
Chem. Soc. Rev., vol. 35, no. 8, pp. 675–83, Aug. 2006.
[23] G. Sandrock, “A panoramic overview of hydrogen storage alloys from a gas reaction point
of view,” J. Alloys Compd., vol. 293–295, pp. 877–888, Dec. 1999.
[24] D. Chung, “Review graphite,” J. Mater. Sci., vol. 37, pp. 1475 – 1489, 2002.
[25] D. Aurbach, E. Zinigrad, Y. Cohen, and H. Teller, “A short review of failure mechanisms
of lithium metal and lithiated graphite anodes in liquid electrolyte solutions,” Solid State
Ionics, vol. 148, pp. 405 – 416, 2002.
[26] M. S. Dresselhaus and G. Dresselhaus, “Intercalation compounds of graphite,” Adv. Phys.,
vol. 51, no. 1, pp. 1–186, Jan. 2002.
[27] W. Lehnert, “The diffusion of lithium through graphite: a Monte Carlo simulation based
on electronic structure calculations,” Chem. Phys., vol. 163, no. 3, pp. 331–337, Jul. 1992.
[28] R. C. Boehm and A. Banerjee, “Theoretical study of lithium intercalated graphite,” J.
Chem. Phys., vol. 96, no. 2, pp. 1150–1157, 1992.
[29] L. Reghai, J. Conard, H. Fuzellier, M. Lelaurain, and E. McRae, “Transport and 13C
nuclear magnetic resonance studies on potassium–ammonia intercalated graphite,” J.
Phys. Chem. Solids, vol. 62, no. 11, pp. 2083–2090, Nov. 2001.
[30] A. Lovell, S. M. Bennington, N. T. Skipper, C. Gejke, H. Thompson, and M. A. Adams,
“Neutron scattering studies of hydrogen in potassium–graphite intercalates: Towards
tunable graphite intercalates for hydrogen storage,” Phys. B Condens. Matter, vol. 385–
386, pp. 163–165, Nov. 2006.
[31] C. Schafhaeutl, “Uber die Verbindungen des Kohlenstoffes mit Silicium, Eisen und
anderen Metallen welche die verschiedenen Gattungen von Roheisen, Stahl und
Schmeideeisen bilden.,” J. Prakt. Chem., vol. 21, pp. 129–157, 1841.
[32] W. Rüdorff and E. Schulze, “Neue Alkali-Graphit-Verbindungen,” Angew. Chem., vol.
66, p. 305, 1954.
143
[33] T. Enoki, M. Suzuki, and M. Endo, Graphite Intercalation Compounds and Applications.
Oxford: Oxford University Press, 2003.
[34] S. A. Solin, “Two-dimensional metal-ammonia-solutions in graphite,” J. Phys. IV, vol. 1,
no. C5, pp. 311–324, 1991.
[35] B. R. York and S. A. Solin, “Effect of composition on charge exchange, lattice expansion,
and staging in potassium-ammonia graphite intercalation compounds,” Phys. Rev. B, vol.
31, no. 12, pp. 8206–8220, 1985.
[36] E. Mayer, “Symmetrical cleavage of diborane by ammonia in solution,” Inorg. Chem., vol.
11, no. 4, pp. 866–869, 1972.
[37] F. H. Stephens, V. Pons, and R. T. Baker, “Ammonia-borane: the hydrogen source par
excellence?,” Dalt. Trans., vol. 2, no. 25, pp. 2613–26, Jul. 2007.
[38] A. D. Sutton, A. K. Burrell, D. a Dixon, E. B. Garner, J. C. Gordon, T. Nakagawa, K. C.
Ott, J. P. Robinson, and M. Vasiliu, “Regeneration of ammonia borane spent fuel by direct
reaction with hydrazine and liquid ammonia.,” Science, vol. 331, no. 6023, pp. 1426–9,
Mar. 2011.
[39] B. Zhong, L. Song, X. Huang, L. Xia, and G. Wen, “First-principles investigation of
ammonia borane for hydrogen storage,” Phys. Scr., vol. 86, p. 015606, 2012.
[40] R. D. Suenram and L. R. Thorne, “Microwave spectrum and dipole moment of NH3BH3,”
Chem. Phys. Lett., vol. 78, no. 1, pp. 1980–1981, 1981.
[41] L. R. Thorne, R. D. Suenram, and F. J. Lovas, “Microwave spectrum, torsional barrier,
and structure of BH3NH3,” J. Chem. Phys., vol. 78, no. 1, pp. 167–171, 1983.
[42] D. West, S. Limpijumnong, and S. Zhang, “Band structures and native defects of ammonia
borane,” Phys. Rev. B, vol. 80, no. 6, pp. 1–10, Aug. 2009.
[43] W. T. Klooster, T. F. Koetzle, P. E. M. Siegbahn, T. B. Richardson, and R. H. Crabtree,
“Study of the N-H•••H-B Dihydrogen Bond Including the Crystal Structure of BH3NH3
by Neutron Diffraction,” J. Am. Chem. Soc., vol. 121, no. 27, pp. 6337–6343, 1999.
[44] M. E. Bowden, G. J. Gainsford, and W. T. Robinson, “Room-Temperature Structure of
Ammonia Borane,” Aust. J. Chem., vol. 60, no. 3, p. 149, 2007.
[45] N. J. Hess, M. E. Bowden, V. M. Parvanov, C. Mundy, S. M. Kathmann, G. K. Schenter,
and T. Autrey, “Spectroscopic studies of the phase transition in ammonia borane: Raman
spectroscopy of single crystal NH3BH3 as a function of temperature from 88 to 330 K.,”
J. Chem. Phys., vol. 128, no. 3, p. 034508, Jan. 2008.
[46] C. R. Miranda and G. Ceder, “Ab initio investigation of ammonia-borane complexes for
hydrogen storage,” J. Chem. Phys., vol. 126, no. 184703, pp. 1–11, 2007.
[47] C. A. Morrison and M. M. Siddick, “Dihydrogen bonds in solid BH3NH3,” Angew. Chem.
Int. Ed. Engl., vol. 43, no. 36, pp. 4780–4782, Sep. 2004.
[48] S. G. Shore and R. W. Parry, “The Crystalline Compound Ammonia-Borane,1 H3NBH3,”
J. Am. Chem. Soc., vol. 77, no. 4, pp. 19–20, 1955.
144
[49] S. G. Shore and K. W. Boddeker, “Large scale synthesis of H2B(NH3)3+BH4
- and
NH3BH3,” Inorg. Chem., vol. 3, no. 6, pp. 914–915, 1964.
[50] P. V. Ramachandran and P. D. Gagare, “Preparation of ammonia borane in high yield and
purity, methanolysis, and regeneration.,” Inorg. Chem., vol. 46, no. 19, pp. 7810–7, Sep.
2007.
[51] C. Reller and F. O. R. L. Mertens, “A self-contained regeneration scheme for spent
ammonia borane based on the catalytic hydrodechlorination of BCl3.,” Angew. Chemie
Int. Ed., vol. 51, no. 47, pp. 11731–5, Nov. 2012.
[52] A. D. Sutton, A. K. Burrell, D. a Dixon, E. B. Garner, J. C. Gordon, T. Nakagawa, K. C.
Ott, J. P. Robinson, and M. Vasiliu, “Regeneration of ammonia borane spent fuel by direct
reaction with hydrazine and liquid ammonia.,” Science, vol. 331, no. 6023, pp. 1426–9,
Mar. 2011.
[53] G. Wolf, J. Baumann, F. Baitalow, and F. P. Hoffmann, “Calorimetric process monitoring
of thermal decomposition of B-N-H compounds,” Thermochim. Acta, vol. 343, pp. 19–25,
2000.
[54] M. Bowden, T. Autrey, I. Brown, and M. Ryan, “The thermal decomposition of ammonia
borane : A potential hydrogen storage material,” Organometallics, vol. 8, pp. 498–500,
2008.
[55] P. A. Storozhenko, R. A. Svitsyn, V. A. Ketsko, A. K. Buryak, and A. V. Ul’yanov,
“Ammineborane: Synthesis and Physicochemical Characterization,” Russ. J. Iorganic
Chem., vol. 50, no. 7, pp. 980–985, 2005.
[56] D. J. Heldebrant, A. Karkamkar, N. J. Hess, M. Bowden, S. Rassat, F. Zheng, K. Rappe,
and T. Autrey, “The Effects of Chemical Additives on the Induction Phase in Solid-State
Thermal Decomposition of Ammonia Borane,” no. 1, pp. 5332–5336, 2008.
[57] S. Frueh, R. Kellett, C. Mallery, T. Molter, W. S. Willis, C. King’ondu, and S. L. Suib,
“Pyrolytic decomposition of ammonia borane to boron nitride.,” Inorg. Chem., vol. 50, no.
3, pp. 783–92, Feb. 2011.
[58] A. Staubitz, A. P. M. Robertson, and I. Manners, “Ammonia-borane and related
compounds as dihydrogen sources.,” Chem. Rev., vol. 110, no. 7, pp. 4079–124, Jul. 2010.
[59] F. Baitalow, J. Baumann, G. Wolf, K. Jaenicke-Rößler, and G. Leitner, “Thermal
decomposition of B–N–H compounds investigated by using combined thermoanalytical
methods,” Thermochim. Acta, vol. 391, no. 1–2, pp. 159–168, Aug. 2002.
[60] J. Zhang, Y. Zhao, D. L. Akins, and J. W. Lee, “Thermal Decomposition and
Spectroscopic Studies of Preheated Ammonia Borane,” J. Phys. Chem. C, vol. 114, no.
45, pp. 19529–19534, Nov. 2010.
[61] V. P. Sorokin, B. I. Vesina, and S. N. Klimova, “Ammonia-Borane: A New Method of
Preparation, and its Properties,” Russ. J. Inorg. Chem., vol. 8, pp. 66–68, 1963.
[62] M. Hu, J. Van Paasschen, and R. Geanangel, “New synthetic approaches to ammonia-
borane and its deuterated derivatives,” J. Inorg. Nucl. Chem., vol. 39, pp. 2147–2150,
1977.
145
[63] R. M. Adams, J. Beres, A. Dodds, and A. J. Morabito, “Dimethyl sulfide-borane as a
borane carrier,” Inorg. Chem., vol. 10, no. 9, pp. 2072–2074, 1971.
[64] S. Sepehri, B. B. Garcia, and G. Cao, “Tuning dehydrogenation temperature of carbon–
ammonia borane nanocomposites,” J. Mater. Chem., vol. 18, no. 34, p. 4034, 2008.
[65] A. C. Gangal, P. Kale, R. Edla, J. Manna, and P. Sharma, “Study of kinetics and thermal
decomposition of ammonia borane in presence of silicon nanoparticles,” Int. J. Hydrogen
Energy, vol. 37, no. 8, pp. 6741–6748, Apr. 2012.
[66] A. C. Stowe, W. J. Shaw, J. C. Linehan, B. Schmid, and T. Autrey, “In situ solid state 11B
MAS-NMR studies of the thermal decomposition of ammonia borane: mechanistic studies
of the hydrogen release pathways from a solid state hydrogen storage material.,” Phys.
Chem. Chem. Phys., vol. 9, no. 15, pp. 1831–6, Apr. 2007.
[67] D. W. Himmelberger, L. R. Alden, M. E. Bluhm, and L. G. Sneddon, “Ammonia borane
hydrogen release in ionic liquids.,” Inorg. Chem., vol. 48, no. 20, pp. 9883–9, Oct. 2009.
[68] D. W. Himmelberger, C. W. Yoon, M. E. Bluhm, P. J. Carroll, and L. G. Sneddon, “Base-
Promoted Ammonia Borane Hydrogen-Release,” ReVision, vol. 6323, no. 1, pp. 73–82,
2009.
[69] V. Sit, R. A. Geanangel, and W. W. Wendlandt, “The thermal dissociation of NH3BH3,”
Thermochim. Acta, vol. 113, pp. 379–382, 1987.
[70] M. Chandra and Q. Xu, “A high-performance hydrogen generation system: Transition
metal-catalyzed dissociation and hydrolysis of ammonia–borane,” J. Power Sources, vol.
156, no. 2, pp. 190–194, Jun. 2006.
[71] M. Chandra and Q. Xu, “Dissociation and hydrolysis of ammonia-borane with solid acids
and carbon dioxide: An efficient hydrogen generation system,” J. Power Sources, vol.
159, no. 2, pp. 855–860, Sep. 2006.
[72] M. Zahmakıran, F. Durap, and S. Özkar, “Zeolite confined copper(0) nanoclusters as cost-
effective and reusable catalyst in hydrogen generation from the hydrolysis of ammonia-
borane,” Int. J. Hydrogen Energy, vol. 35, no. 1, pp. 187–197, Jan. 2010.
[73] K. W. Boddeker, S. G. Shore, and R. K. Bunting, “Boron-Nitrogen Chemistry.,” J. Am.
Chem. Soc., vol. 2695, no. 9, pp. 4396–4401, 1966.
[74] D. W. Himmelberger, “Hydrogen release form ammonia borane,” 2010.
[75] H. C. Kelly and V. B. Mattiott, “Reexamination of the acid-catalysed amine-borane
hydrolysis. The hydrolysis of NH3•BH3,” Inorg. Chem., vol. 18, no. 10, pp. 2875–2878,
1979.
[76] M. Chandra and Q. Xu, “Room temperature hydrogen generation from aqueous ammonia-
borane using noble metal nano-clusters as highly active catalysts,” J. Power Sources, vol.
168, no. 1, pp. 135–142, May 2007.
[77] T. Wideman and L. G. Sneddon, “Convenient Procedures for the Laboratory Preparation
of Borazine,” Inorg. Chem., vol. 34, no. 4, pp. 1002–1003, Feb. 1995.
146
[78] W. J. Shaw, J. C. Linehan, N. K. Szymczak, D. J. Heldebrant, C. Yonker, D. M. Camaioni,
R. T. Baker, and T. Autrey, “In situ multinuclear NMR spectroscopic studies of the
thermal decomposition of ammonia borane in solution.,” Angew. Chem. Int. Ed. Engl., vol.
47, no. 39, pp. 7493–6, Jan. 2008.
[79] J. S. Wang and R. A. Geanangel, “11B NMR Studies of the Thermal Decomposition of
Ammonia-Borane in Solution,” Inorganica Chim. Acta, vol. 148, pp. 185–190, 1988.
[80] Y. Kim, H. Baek, J. H. Lee, S. Yeo, K. Kim, S.-J. Hwang, B. Eun, S. W. Nam, T.-H. Lim,
and C. W. Yoon, “Metal-free, polyether-mediated H2-release from ammonia borane: roles
of hydrogen bonding interactions in promoting dehydrogenation.,” Phys. Chem. Chem.
Phys., vol. 15, no. 45, pp. 19584–94, Dec. 2013.
[81] H. M. Colquhoun, G. Jones, J. M. Maud, J. F. Stoddart, and D. J. Williams, “Crystal and
supramolecular strucutres of complexes* of BF3NH3 and BH3NH3 with 18-crown-6,” J.
Chem. Soc. Dalt. Trans., pp. 147–149, 1984.
[82] M. E. Bluhm, M. G. Bradley, R. I. Butterick, U. Kusari, and L. G. Sneddon,
“Amineborane-based chemical hydrogen storage: enhanced ammonia borane
dehydrogenation in ionic liquids,” J. Am. Chem. Soc., vol. 3, pp. 7748–7749, 2006.
[83] W. R. H. Wright, E. R. Berkeley, L. R. Alden, R. T. Baker, and L. G. Sneddon, “Transition
metal catalysed ammonia-borane dehydrogenation in ionic liquids.,” Chem. Commun.,
vol. 47, no. 11, pp. 3177–9, Mar. 2011.
[84] A. Feaver, S. Sepehri, P. Shamberger, A. Stowe, T. Autrey, and G. Cao, “Coherent carbon
cryogel-ammonia borane nanocomposites for H2 storage.,” J. Phys. Chem. B, vol. 111,
no. 26, pp. 7469–72, Jul. 2007.
[85] T. Zhang, X. Yang, S. Yang, D. Li, F. Cheng, Z. Tao, and J. Chen, “Silica hollow
nanospheres as new nanoscaffold materials to enhance hydrogen releasing from ammonia
borane,” Phys. Chem. Chem. Phys., 2011.
[86] A. C. Gangal, R. Edla, K. Iyer, R. Biniwale, M. Vashistha, and P. Sharma, “Effect of
zeolites on thermal decomposition of ammonia borane,” Int. J. Hydrogen Energy, vol. 37,
no. 4, pp. 3712–3718, Feb. 2012.
[87] S. Sepehri and A. Feaver, “Spectroscopic studies of dehydrogenation of ammonia borane
in carbon cryogel,” J. Phys. Chem. B, vol. 111, pp. 14285–14289, 2007.
[88] K. Chang, E. Kim, P. F. Weck, and D. Tománek, “Nanoconfinement effects on the
reversibility of hydrogen storage in ammonia borane: a first-principles study.,” J. Chem.
Phys., vol. 134, no. 21, p. 214501, Jun. 2011.
[89] Z. Tang, H. Chen, and X. Chen, “Graphene Oxide Based Recyclable Dehydrogenation of
Ammonia Borane within a Hybrid Nanostructure,” J. Am. …, 2012.
[90] L. Li, X. Yao, C. Sun, A. Du, L. Cheng, Z. Zhu, C. Yu, J. Zou, S. C. Smith, P. Wang, H.-
M. Cheng, R. L. Frost, and G. Q. (Max) Lu, “Lithium-Catalyzed Dehydrogenation of
Ammonia Borane within Mesoporous Carbon Framework for Chemical Hydrogen
Storage,” Adv. Funct. Mater., vol. 19, no. 2, pp. 265–271, Jan. 2009.
147
[91] S. B. Kalidindi, J. Joseph, and B. R. Jagirdar, “Cu2+-induced room temperature hydrogen
release from ammonia borane,” Energy Environ. Sci., vol. 2, no. 12, p. 1274, 2009.
[92] T. He, Z. Xiong, G. Wu, H. Chu, C. Wu, T. Zhang, and P. Chen, “Nanosized Co- and Ni-
Catalyzed Ammonia Borane for Hydrogen Storage,” Chem. Mater., vol. 21, no. 11, pp.
2315–2318, Jun. 2009.
[93] X. Kang, Z. Fang, L. Kong, H. Cheng, X. Yao, G. Lu, and P. Wang, “Ammonia Borane
Destabilized by Lithium Hydride: An Advanced On-Board Hydrogen Storage Material,”
Adv. Mater., vol. 20, no. 14, pp. 2756–2759, Jul. 2008.
[94] Z. Xiong, C. K. Yong, G. Wu, P. Chen, W. Shaw, A. Karkamkar, T. Autrey, M. O. Jones,
S. R. Johnson, P. P. Edwards, and W. I. F. David, “High-capacity hydrogen storage in
lithium and sodium amidoboranes.,” Nat. Mater., vol. 7, no. 2, pp. 138–41, Feb. 2008.
[95] A. T. Luedtke and T. Autrey, “Hydrogen release studies of alkali metal amidoboranes.,”
Inorg. Chem., vol. 49, no. 8, pp. 3905–10, Apr. 2010.
[96] J. Luo, X. Kang, and P. Wang, “Synthesis, formation mechanism, and dehydrogenation
properties of the long-sought Mg(NH2BH3)2compound,” Energy Environ. Sci., vol. 6, no.
3, p. 1018, 2013.
[97] Z. Tang, S. Li, Z. Yang, and X. Yu, “Ammonia borane nanofibers supported by poly(vinyl
pyrrolidone) for dehydrogenation,” J. Mater. Chem., vol. 21, pp. 14616–14621, 2011.
[98] S. F. Li, Z. W. Tang, Y. B. Tan, and X. B. Yu, “Polyacrylamide Blending with Ammonia
Borane: A Polymer Supported Hydrogen Storage Composite,” J. Phys. Chem. C, vol. 116,
no. 1, pp. 1544–1549, Jan. 2012.
[99] J. Zhao, J. Shi, X. Zhang, F. Cheng, J. Liang, Z. Tao, and J. Chen, “A soft hydrogen storage
material: poly(methyl acrylate)-confined ammonia borane with controllable
dehydrogenation.,” Adv. Mater., vol. 22, no. 3, pp. 394–7, Jan. 2010.
[100] Z. Kurban, A. Lovell, S. M. Bennington, D. W. K. Jenkins, K. R. Ryan, M. O. Jones, N.
T. Skipper, and W. I. F. David, “A Solution Selection Model for Coaxial Electrospinning
and Its Application to Nanostructured Hydrogen Storage Materials,” J. Phys. Chem. C, no.
114, pp. 21201–21213, 2010.
[101] Yuan-Huffman, Q, W, “Poly(ethylene) oxide,” in Polymer data handbook, 2nd ed.,
Oxford: Oxford University Press, 2009, pp. 690–700.
[102] C. Vasile and A. K. Kulshreshtha, Handbook of polymer blends and composites, Volumes
1 - 4. Shrewsbury: Rapra Technology, 2002.
[103] J. E. Mark, Polymer Data Handbook, 2nd ed. Oxford: Oxford University Press, 2009.
[104] R. J. Keaton, J. M. Blacquiere, and R. T. Baker, “Base Metal Catalyzed Dehydrogenation
of Ammonia-Borane for Chemical Hydrogen Storage,” vol. 2, pp. 1844–1845, 2006.
[105] A. Gutowska, L. Li, Y. Shin, C. M. Wang, X. S. Li, J. C. Linehan, R. S. Smith, B. D. Kay,
B. Schmid, W. Shaw, M. Gutowski, and T. Autrey, “Nanoscaffold Mediates Hydrogen
Release and the Reactivity of Ammonia Borane,” Angew. Chem. Int. Ed., pp. 3578 –3582,
2005.
148
[106] J. J. Vajo and G. L. Olson, “Hydrogen storage in destabilized chemical systems,” Scr.
Mater., vol. 56, no. 10, pp. 829–834, May 2007.
[107] S. Ramakrishna, K. Fujihara, W. Teo, T. Lim, and Z. Ma, An Introduction to
Electrospinning and Nanofibers. Sibgapore: world scientific Publishing Co. Pte. Ltd.,
2005.
[108] Y. Pan, Y. Wang, Y. Liang, Z. Tao, and J. Chen, “Promoted hydrogen release from
ammonia borane with mannitol via a solid-state reaction route.,” Dalton Trans., vol. 41,
no. 3, pp. 871–5, Jan. 2012.
[109] W. A. Deer, R. A. Howie, and J. Zussman, An introduction to the rock-forming minerals,
2nd ed. Harlow, Essex: Longman Scientific and Technical, 1992.
[110] B. Ruzicka and E. Zaccarelli, “A fresh look at the Laponite phase diagram,” Soft Matter,
vol. 7, no. 4, p. 1268, 2011.
[111] C. Zanzottera, A. Vicente, E. Celasco, C. Fernandez, E. Garrone, and B. Bonelli, “Physico-
Chemical Properties of Imogolite Nanotubes Functionalized on Both External and Internal
Surfaces,” J. Phys. Chem. C, vol. 116, no. 13, pp. 7499–7506, Apr. 2012.
[112] J. P. Gabriel and P. Davidson, “Mineral Liquid Crystals from Self-Assembly of
Anisotropic Nanosystems,” 2003, pp. 119–172.
[113] B. Creton, D. Bougeard, K. S. Smirnov, J. Guilment, and O. V. Poncelet, “Molecular
dynamics study of hydrated imogolite,” Phys. Chem. Chem. Phys., vol. 10, no. 32, pp.
4879–4888, Aug. 2008.
[114] P. D. G. Cradwick, V. C. Farmer, and J. D. Russell, “Imogolite, a hydrated aluminium
silicate of tubular structure,” Nature, vol. 240, pp. 187–187, 1972.
[115] W. Ma, W. O. Yah, H. Otsuka, and A. Takahara, “Application of imogolite clay nanotubes
in organic–inorganic nanohybrid materials,” J. Mater. Chem., vol. 22, no. 24, p. 11887,
2012.
[116] E. Joussein, S. Petit, J. Churchman, and B. Theng, “Halloysite clay minerals—a review,”
Clay Miner., vol. 40, pp. 383–426, 2005.
[117] R. Kamble, M. Ghag, S. Gaikawad, and B. K. Panda, “Review Article Halloysite
Nanotubes and Applications : A Review,” vol. 3, no. 2, pp. 25–29, 2012.
[118] J. K. Walters, N. T. Skipper, and A. K. Soper, “The interlayer structure of a graphite–
potassium–ammonia intercalation compound by neutron diffraction,” Chem. Phys. Lett.,
vol. 300, no. February, pp. 444–450, 1999.
[119] H. Zabel, “Dynamics of molecules in graphite,” Synth. Met., vol. 23, pp. 37–42, 1988.
[120] F. Leclercq, P. Damay, and P. Chieux, “Quantum and Diffusional Rotation of Ammonia
in Hexaamminecalcium As Observed by Incoherent Neutron Scattering,” J. Phys. Chem.,
vol. 88, pp. 3886–3890, 1984.
[121] H. Zabel and D. A. Neumann, “Neutron scattering studies of potassium-ammonia layers
in graphite,” Can. J. Chem., vol. 66, pp. 666–671, 1988.
149
[122] H. A. Resing, B. R. York, S. A. Solin, and R. M. Fronko, “Preferred Orientation of
Ammonia in KC24(NH3)4.3 Found by Proton NMR Spectroscopy,” in 17th Biennial
Conference at University of Kentucky, Lexington, Kentucky, 1985, vol. 24, pp. 4–5.
[123] T. Tsang, R. M. Fronko, H. A. Resing, X. W. Qain, and S. A. Solin, “1H NMR Study of
Ternary Ammonia-Alkali Metal-Graphite Intercalation Compounds.,” Solid State
Commun., vol. 63, no. 4, pp. 361–366, 1987.
[124] Q. P. Pham, U. Sharma, and A. G. Mikos, “Electrospinning of Polymeric Nanofibers for
Tissue Engineering Applications : A Review,” Tissue Eng., vol. 12, no. 5, 2006.
[125] Y. Chen, Z. Lu, L. Zhou, Y.-W. Mai, and H. Huang, “In situ formation of hollow graphitic
carbon nanospheres in electrospun amorphous carbon nanofibers for high-performance Li-
based batteries.,” Nanoscale, vol. 4, no. 21, pp. 6800–5, Nov. 2012.
[126] J. S. Choi, Y. Kim, J. Kang, S. Y. Jeong, and H. S. Yoo, “Electrospun chitosan
microspheres for complete encapsulation of anionic proteins: controlling particle size and
encapsulation efficiency.,” AAPS PharmSciTech, vol. 14, no. 2, pp. 794–801, Jun. 2013.
[127] B. a. Miller-Chou and J. L. Koenig, “A review of polymer dissolution,” Prog. Polym. Sci.,
vol. 28, no. 8, pp. 1223–1270, Aug. 2003.
[128] T. L. Yu, “Polymer Molecular Weight from Loss Modulus.,” Polym. J., vol. 24, no. 7, pp.
220–1328, 1992.
[129] S. Ramakrishna, K. Fujihara, W. E. Teo, T. C. Lim, and Z. Ma, An Introduction to
Electrospinning and Nanofibres. Singapore: world scientific Publishing Co. Pte. Ltd.,
2005.
[130] Y. Liu, J. He, J. Yu, and H. Zeng, “Controlling numbers and sizes of beads in electrospun
nanofibers,” Polym. Int., vol. 636, no. August 2007, pp. 632–636, 2008.
[131] G. Eda and S. Shivkumar, “Bead‐ to‐ fiber transition in electrospun polystyrene,” J. Appl.
Polym. Sci., no. May 2006, 2007.
[132] Z.-M. Huang, Y.-Z. Zhang, M. Kotaki, and S. Ramakrishna, “A review on polymer
nanofibers by electrospinning and their applications in nanocomposites,” Compos. Sci.
Technol., vol. 63, no. 15, pp. 2223–2253, Nov. 2003.
[133] C. J. Luo, M. Nangrejo, and M. Edirisinghe, “A novel method of selecting solvents for
polymer electrospinning,” Polymer (Guildf)., vol. 51, no. 7, pp. 1654–1662, Mar. 2010.
[134] S. Koombhongse, W. Liu, and D. H. Reneker, “Flat Polymer Ribbons and Other Shapes
by Electrospinning,” Polymer (Guildf)., no. May, pp. 2598–2606, 2001.
[135] C. J. Angammana, S. Member, and S. H. Jayaram, “Analysis of the Effects of Solution
Conductivity on Electrospinning Process and Fiber Morphology,” vol. 47, no. 3, pp. 1109–
1117, 2011.
[136] G. H. Lee, J. C. Song, and K. B. Yoon, “Controlled wall thickness and porosity of
polymeric hollow nanofibers by coaxial electrospinning,” Macromol. Res., vol. 18, no. 6,
pp. 571–576, Jun. 2010.
150
[137] K. H. Lee, H. Y. Kim, H. J. Bang, Y. H. Jung, and S. G. Lee, “The change of bead
morphology formed on electrospun polystyrene fibers,” Polymer (Guildf)., vol. 44, no. 14,
pp. 4029–4034, Jun. 2003.
[138] X. Zong, K. Kim, D. Fang, S. Ran, B. S. Hsiao, and B. Chu, “Structure and process
relationship of electrospun bioabsorbable nanofiber membranes,” Polymer (Guildf)., vol.
43, pp. 4403–4412, 2002.
[139] C. Wang, C.-H. Hsu, and J.-H. Lin, “Scaling Laws in Electrospinning of Polystyrene
Solutions,” Macromolecules, vol. 39, no. 22, pp. 7662–7672, Oct. 2006.
[140] G. Larsen, R. Spretz, and R. Velarde-Ortiz, “Use of Coaxial Gas Jackets to Stabilize
Taylor Cones of Volatile Solutions and to Induce Particle-to-Fiber Transitions,” Adv.
Mater., vol. 16, no. 2, pp. 166–169, Jan. 2004.
[141] J. A. Nairn, “Viscosity,” 2003.
[142] P. Sunthar, “Polymer Rheology,” in Rheology of Complex Fluids, New York: Springer
New York, 2010, pp. 171–191.
[143] J.-H. He, Y.-Q. Wan, and J.-Y. Yu, “Effect of concentration on electrospun
polyacrylonitrile (PAN) nanofibers,” Fibers Polym., vol. 9, no. 2, pp. 140–142, Jul. 2008.
[144] M. M. Demir, “Investigation on glassy skin formation of porous polystyrene fibers
electrospun from DMF,” eXPRESS Polym. Lett., vol. 4, no. 1, pp. 2–8, Dec. 2009.
[145] K. A. Dill and S. Bromberg, Molecular Driving Forces: Statistical Thermodynamics in
Chemistry and Biology. London: Garland Science, 2003.
[146] D. H. Reneker, A. L. Yarin, E. Zussman, and H. Xu, “Electrospinning of Nanofibers from
Polymer Solutions and Melts,” Adv. Appl. Mech., vol. 41, p. 43, 2007.
[147] E. Antoniou and P. Alexandridis, “Polymer conformation in mixed aqueous-polar organic
solvents,” Eur. Polym. J., vol. 46, no. 2, pp. 324–335, Feb. 2010.
[148] G.-W. Oetjen, “Freeze-Drying,” Ullmann’s Encyclopedia of Industrial Chemistry. Wiley-
VCH Verlag GmbH & Co, pp. 52–100, 2012.
[149] G. Srinivas, C. A. Howard, S. M. Bennington, N. T. Skipper, and M. Ellerby, “Effect of
hydrogenation on structure and superconducting properties of CaC6,” J. Mater. Chem.,
vol. 19, no. 29, p. 5239, 2009.
[150] A. Kriete, H. Gundlach, S. Amelinckx, and L. Reimer, “Microscopy,” Ullmann’s
Encyclopedia of Industrial Chemistry. Wiley-VCH Verlag GmbH & Co, pp. 298–308,
2012.
[151] J. Coates, “Interpretation of Infrared Spectra , A Practical Approach,” Encyclopedia of
Analytical Chemistry. John Wiley& Sons Ltd, Chichester, pp. 10815–10837, 2000.
[152] Y.-C. Ning, “Interpretation of Infrared Spectra,” in Interpretation of Organic Spectra,
Singapore: John Wiley & Sons (Asia) Pte Ltd, 2011, pp. 129–146.
151
[153] H.-U. Remlich, “Infrared and Raman Spectroscopy,” Ullmann’s Encyclopedia of
Industrial Chemistry. pp. 177–220, 2012.
[154] a Cooper and C. M. Johnson, “Differential scanning calorimetry.,” Methods Mol. Biol.,
vol. 22, pp. 125–36, Jan. 1994.
[155] S. B. Warrington and G. W. H. Hohne, “Thermal Analysis and Calorimetry,” Ullmann’s
Encyclopedia of Industrial Chemistry. WILEY-VCH Verlag GmbH & Co. KGaA, pp.
415–439, 2012.
[156] J. H. Gross, Mass spectrometry : a textbook. Berlin: Springer, 2001.
[157] A. Savitzky and M. J. E. Golay, “Smoothing and differentialtion of data by simplified least
squares procedures,” J. Anal. Chem., vol. 36, no. 8, pp. 1627–1639, 1964.
[158] G. Orwell, Animal Farm: A Fairy Story, New Ed. London, UK: Penguin Classics, 2010.
[159] K. N. Ninan, “Kinetics of solid state termal decomposition reactions,” J. Therm. Anal.,
vol. 35, pp. 1267–1278, 1989.
[160] A. C. Gangal, P. Kale, R. Edla, J. Manna, and P. Sharma, “Study of kinetics and thermal
decomposition of ammonia borane in presence of silicon nanoparticles,” Int. J. Hydrogen
Energy, vol. 37, no. 8, pp. 6741–6748, Apr. 2012.
[161] D. S. Sivia, Elementary Scattering Theory, For X-ray and Neutron Users. Oxford: Oxford
University Press, 2011.
[162] Y. Waseda, E. Matsubara, and K. Shinoda, X-ray diffraction crystallography:
introduction, examples and solved problems. Berlin: Springer, 2011, pp. 1–310.
[163] Fatima Gonzales Sanchez, “Water diffusion through compacted clays analyzed by neutron
scattering and tracer experiments,” 2007.
[164] A. Lovell, “Tuneable graphite intercalates for hydrogen storage,” 2007.
[165] B. Willis and C. J. Carlile, Experimental Neutron Scattering. Oxford: Oxford University
Press, 2009, pp. 295–309.
[166] K. S. Singwi and A. Sjölander, “Diffusive motions in water and cold neutron scattering,”
Phys. Rev., vol. 119, no. 3, pp. 863–871, 1960.
[167] P. L. Hall and D. K. Ross, “Incoherent neutron scattering functions for random jump
diffusion in bounded and infinite media,” Mol. Phys. An Int. J. Interface Between Chem.
Phys., vol. 42, no. 3, pp. 673–682, 1981.
[168] C. T. Chudley and R. J. Elliott, “Neutron scattering from a liquid on a jump diffusion
model,” Proc. Phys. Soc., vol. 353, no. 77, pp. 353–361, 1961.
[169] W. S. Howells, V. García Sakai, F. Demmel, M. T. F. Telling, and F. Fernandez-Alonso,
“The MODES User Guide v3,” Analysis, no. February. Rutherford Appleton Laboratory
Technical Report, pp. 1–62, 2010.
152
[170] W. E. Teo and S. Ramakrishna, “A review on electrospinning design and nanofibre
assemblies.,” Nanotechnology, vol. 17, no. 14, pp. R89–R106, Jul. 2006.
[171] J. Deitzel, “Controlled deposition of electrospun poly(ethylene oxide) fibers,” Polymer
(Guildf)., vol. 42, no. 19, pp. 8163–8170, Sep. 2001.
[172] H. Tadokoro, Y. Chatani, T. Yoshihara, S. Tahara, and S. Murahashi, “Structural Studies
on Polyethers, [-(CH2)m-O-]n. II. Molecular Structure of Polyethylene Oxide,” pp. 109–
127, 1963.
[173] F. Gu, H. Bu, and Z. Zhang, “A unique morphology of freeze-dried poly(ethylene oxide)
and its transformation,” Polymer (Guildf)., vol. 41, no. 21, pp. 7605–7609, Oct. 2000.
[174] A. C. French, A. L. Thompson, and B. G. Davis, “High-purity discrete PEG-oligomer
crystals allow structural insight.,” Angew. Chem. Int. Ed. Engl., vol. 48, no. 7, pp. 1248–
52, Jan. 2009.
[175] Mettler Toledo, “Interpreting DSC curves Part 1: Dynamic measurements,” 2000.
[176] D. Wolstenholme and K. Traboulsee, “Thermal desorption of hydrogen from ammonia
borane: unexpected role of homopolar B–H···H–B interactions,” Chem. Commun., vol.
48, pp. 2597–2599, 2012.
[177] F. Merten, G. Wolf, and F. Baitalow, “Ammonia borane and related compouds as
hydrogen source materials,” in Handbook of Hydrogen Storage New Materails for Future
Energy Storage, M. Hirscher, Ed. Weinheim: Wiley-VCH Verlag GmbH & Co, 2010, pp.
215–244.
[178] J. Zhang, Y. Zhao, D. Akins, and J. Lee, “CO2-enhanced thermolytic H2 release from
ammonia borane,” J. Phys. …, vol. 115, pp. 8386–8392, 2011.
[179] U. B. Demirci, S. Bernard, R. Chiriac, F. Toche, and P. Miele, “Hydrogen release by
thermolysis of ammonia borane NH3BH3 and then hydrolysis of its by-product [BNHx],”
J. Power Sources, vol. 196, no. 1, pp. 279–286, Jan. 2011.
[180] S. Xie, Y. Song, and Z. Liu, “In situ high-pressure study of ammonia borane by Raman
and IR spectroscopy,” Can. J. Chem., vol. 87, no. 9, pp. 1235–1247, Sep. 2009.
[181] N. Gondaliya, Dinesh Kumar Kanchan, P. Sharma, and P. Joge, “Structural and
Conductivity Studies of Poly(Ethylene Oxide) – Silver Triflate Polymer Electrolyte
System,” Mater. Sci. Appl., vol. 2, pp. 1639–1643, 2011.
[182] M. Sundar and S. Selladurai, “Effect of fillers on magnesium–poly(ethylene oxide) solid
polymer electrolyte,” Ionics (Kiel)., vol. 12, no. 4–5, pp. 281–286, Oct. 2006.
[183] J. Joseph and E. D. Jemmis, “Red-, blue-, or no-shift in hydrogen bonds: a unified
explanation.,” J. Am. Chem. Soc., vol. 129, no. 15, pp. 4620–32, Apr. 2007.
[184] L. Jun, X. Shuping, and G. Shiyang, “FT-IR and Raman spectroscopic study of hydrated
borates,” Spectrochim. Acta Part A Mol. Biomol. Spectrosc., vol. 51, no. 4, pp. 519–532,
Apr. 1995.
153
[185] S. Gadipelli, J. Ford, W. Zhou, H. Wu, T. J. Udovic, and T. Yildirim, “Nanoconfinement
and Catalytic Dehydrogenation of Ammonia Borane by Magnesium-Metal–Organic-
Framework-74,” Communication, pp. 6043 – 6047, 2011.
[186] A. Kanturk, M. Sari, and S. Piskin, “Synthesis , crystal structure and dehydration kinetics
of NaB ( OH ) 4 · 2H 2 O,” vol. 25, no. 6, pp. 1331–1337, 2008.
[187] A. Y. Likhacheva, E. A. Paukshtis, Y. V. Seryotkin, and S. G. Shulgenko, “IR
spectroscopic characterization of NH4-analcime,” Phys. Chem. Miner., vol. 29, no. 9, pp.
617–623, Oct. 2002.
[188] R. C. Taylor and C. L. Cluff, “Vibrational Frequency associated with the Boron-Nitrogen
Dative Bond in Amine Boranes,” Nature, vol. 182, pp. 390–391, 1958.
[189] B. Kang and G. Ceder, “Battery materials for ultrafast charging and discharging.,” Nature,
vol. 458, no. 7235, pp. 190–3, Mar. 2009.
[190] D.-P. Kim, K.-T. Moon, J.-G. Kho, J. Economy, C. Gervais, and F. Babonneau, “Synthesis
and characterization of poly(aminoborane) as a new boron nitride precursor,” Polym. Adv.
Technol., vol. 10, no. 12, pp. 702–712, Dec. 1999.
[191] J. F. Kostka, R. Schellenberg, F. Baitalow, T. Smolinka, and F. Mertens, “Concentration-
Dependent Dehydrogenation of Ammonia-Borane/Triglyme Mixtures,” Eur. J. Inorg.
Chem., vol. 2012, no. 1, pp. 49–54, Jan. 2012.
[192] C. L. Casper, J. S. Stephens, N. G. Tassi, D. B. Chase, and J. F. Rabolt, “Controlling
Surface Morphology of Electrospun Polystyrene Fibers: Effect of Humidity and Molecular
Weight in the Electrospinning Process,” Macromolecules, vol. 37, no. 2, pp. 573–578, Jan.
2004.
[193] S. Megelski, J. S. Stephens, D. B. Chase, and J. F. Rabolt, “Micro- and Nanostructured
Surface Morphology on Electrospun Polymer Fibers,” Macromolecules, vol. 35, no. 22,
pp. 8456–8466, Oct. 2002.
[194] T. Jarusuwannapoom, W. Hongrojjanawiwat, S. Jitjaicham, L. Wannatong, M.
Nithitanakul, C. Pattamaprom, P. Koombhongse, R. Rangkupan, and P. Supaphol, “Effect
of solvents on electro-spinnability of polystyrene solutions and morphological appearance
of resulting electrospun polystyrene fibers,” Eur. Polym. J., vol. 41, no. 3, pp. 409–421,
Mar. 2005.
[195] G. C. Andrews and S. F. Neelamkavil, “Borane–Ammonia,” E-EROS Encyclopedia of
Reagents for Organic Synthesis, no. eq 5. pp. 1–2, 2008.
[196] G. Eda, J. Liu, and S. Shivkumar, “Solvent effects on jet evolution during electrospinning
of semi-dilute polystyrene solutions,” Eur. Polym. J., vol. 43, no. 4, pp. 1154–1167, Apr.
2007.
[197] E. S. Medeiros, L. H. C. Mattoso, R. D. Offeman, D. F. Wood, and W. J. Orts, “Effect of
relative humidity on the morphology of electrospun polymer fibers,” Can. J. Chem., vol.
86, no. 6, pp. 590–599, 2008.
[198] A. C. D. Newman, Chemsitry of Clays and Clay Minerals. Harlow, Essex: Longman
Scientific and Technical, 1987.
154
[199] V. C. Farmer, M. J. Adams, A. R. Fraser, and F. Palmieri, “Synthetic imogolite: properties,
synthesis, and possible applications,” Clay Miner., vol. 18, pp. 459–472, 1983.
[200] K. Wada and N. Yoshinaga, “The structure of ‘imogolite,’” Am. Mineral., vol. 54, pp. 50–
71, 1969.
[201] C. Levard, A. Masion, J. Rose, E. Doelsch, D. Borschneck, C. Dominici, F. Ziarelli, and
J.-Y. Bottero, “Synthesis of imogolite fibers from decimolar concentration at low
temperature and ambient pressure: a promising route for inexpensive nanotubes.,” J. Am.
Chem. Soc., vol. 131, no. 47, pp. 17080–1, Dec. 2009.
[202] G. H. Koenderink, S. G. J. M. Kluijtmans, and A. P. Philipse, “On the synthesis of
colloidal imogolite fibers,” J. Colloid Interface Sci., vol. 216, pp. 429–431, 1999.
[203] D.-Y. Kang, J. Zang, E. R. Wright, A. L. McCanna, C. W. Jones, and S. Nair,
“Dehydration, dehydroxylation, and rehydroxylation of single-walled aluminosilicate
nanotubes.,” ACS Nano, vol. 4, no. 8, pp. 4897–907, Aug. 2010.
[204] B. Bonelli, I. Bottero, N. Ballarini, S. Passeri, F. Cavani, and E. Garrone, “IR
spectroscopic and catalytic characterization of the acidity of imogolite-based systems,” J.
Catal., vol. 264, no. 1, pp. 15–30, May 2009.
[205] D. Cheddie, “Ammonia as a Hydrogen Source for Fuel Cells: A Review,” in Hydrogen
Energy - Challenges and Perspectives, D. Minic, Ed. InTech, 2012.
[206] G. Srinivas, a Lovell, N. T. Skipper, S. M. Bennington, Z. Kurban, and R. I. Smith,
“Ammonia absorption in calcium graphite intercalation compound: in situ neutron
diffraction, Raman spectroscopy and magnetization.,” Phys. Chem. Chem. Phys., vol. 12,
no. 23, pp. 6253–9, Jun. 2010.
[207] M. Hӧhne, Y. X. Wang, and E. Stumpp, “New ternary alkaline earth metal-ammonia
compounds,” Synth. Met., vol. 34, pp. 41–46, 1989.
[208] H. A. Resing, B. R. York, S. A. Solin, and R. M. Fronko, “Preferred Orientation of
Ammonia in KC24(NH3) Found by Proton NMR Spectroscopy,” 17th Bienn. Conf. Univ.
Kentucky, Lexington, Kentucky, vol. 24, pp. 4–5, 1985.
[209] G. Srinivas, A. Lovell, N. T. Skipper, S. M. Bennington, Z. Kurban, and R. I. Smith,
“Ammonia absorption in calcium graphite intercalation compound: in situ neutron di
raction, Raman spectroscopy and magnetization,” Phys. Chem. Chem. Phys., vol. 12, pp.
6253–6259, 2010.
[210] C. J. Carlile, I. M. Jamie, G. Lockhart, and J. W. White, “Two dimensional caesium-
ammonia solid sollution in C228Cs(NH3)x,” Mol. Phys., vol. 76, no. 1, pp. 173–200, 1992.
[211] D. A. Neumann, H. Zabel, J. J. Rush, Y. B. Fan, and S. A. Solin, “Quasi-elastic neutron
scattering study of rotations and diffusion in KC24(NH3)4.3,” J. Phys. Chem. C, vol. 761,
no. 20, pp. L761–L764, 1987.
[212] D. S. Kyser and J. C. Thompson, “Measurement of the Hall Effect in Metal—Ammonia
Solutions,” J. Chem. Phys., vol. 42, no. 11, p. 3910, 1965.
155